Add release v2.1.3

Fix release script for plugin latest_release

.github/ISSUE_TEMPLATE.md

@@ -0,0 +1,65 @@
+<!--
+Hi, thank you for opening an issue!
+Before hitting the button...
+
+** Is this a REQUEST FOR HELP? **
+If so, please have a look at:
+- our FAQ: https://www.weave.works/docs/net/latest/faq/
+- our troubleshooting page: https://www.weave.works/docs/net/latest/troubleshooting/
+- our help page, to choose the best channel (Slack, etc.) to reach out: https://www.weave.works/help/
+
+** Is this a FEATURE REQUEST? **
+If so, please search existing feature requests, and if you find a similar one, up-vote it and/or add your comments to it instead.
+If you did not find a similar one, please describe in details:
+- why: your use-case, specific constraints you may have, etc.
+- what: the feature/behaviour/change you would like to see in Weave Net
+Do not hesitate, when appropriate, to share the exact commands or API you would like, and/or to share a diagram (e.g.: asciiflow.com): "a picture is worth a thousand words".
+
+** Is this a BUG REPORT? **
+Please fill in as much of the template below as you can.
+
+Thank you!
+-->
+
+## What you expected to happen?
+
+## What happened?
+<!-- Error message, actual behaviour, etc. -->
+
+## How to reproduce it?
+<!-- Specific steps, as minimally and precisely as possible. -->
+
+## Anything else we need to know?
+<!-- Cloud provider? Hardware? How did you configure your cluster? Kubernetes YAML, KOPS, etc. -->
+
+## Versions:
+<!-- Please paste in the output of these commands; 'kubectl' only if using Kubernetes -->
+```
+$ weave version
+$ docker version
+$ uname -a
+$ kubectl version
+```
+
+## Logs:
+```
+$ docker logs weave
+```
+or, if using Kubernetes:
+```
+$ kubectl logs -n kube-system <weave-net-pod> weave
+```
+<!-- (If output is long, please consider a Gist.) -->
+<!-- Anything interesting or unusual output by the below, potentially relevant, commands?
+$ journalctl -u docker.service --no-pager
+$ journalctl -u kubelet --no-pager
+$ kubectl get events
+-->
+
+## Network:
+<!-- If your problem has anything to do with one network endpoint not being able to contact another, please run the following commands -->
+```
+$ ip route
+$ ip -4 -o addr
+$ sudo iptables-save
+```

.gitignore

@@ -33,11 +33,13 @@ coverage.html
 # Vim backup files
 *.swp
 
+# Sed backup files
+*.bak
+
 # Project specific
 prog/weaver/weaver
 prog/weavedns/weavedns
 prog/weavehosts/weavehosts
-prog/weaveproxy/weaveproxy
 prog/sigproxy/sigproxy
 prog/weavewait/weavewait
 prog/weavewait/weavewait_noop
@@ -53,14 +55,13 @@ testing/runner/runner
 releases
 .*.uptodate
 weave*.tar.gz
-prog/weaveexec/weave
+prog/weaver/weave
 prog/weaveexec/weavehosts
-prog/weaveexec/weaveproxy
 prog/weaveexec/weavewait
 prog/weaveexec/weavewait_noop
 prog/weaveexec/weavewait_nomcast
 prog/weaveexec/sigproxy
-prog/weaveexec/weaveutil
+prog/weaver/weaveutil
 prog/weaveexec/docker*
 Vagrantfile.local
 test/tls/tls
@@ -70,6 +71,7 @@ test/coverage.*
 *qemu-*-static
 terraform.tfstate
 terraform.tfstate.backup
+prog/net-plugin/rootfs/
 
 # Website
 site-build

.gitmodules

@@ -124,3 +124,57 @@
 [submodule "vendor/github.com/containernetworking/cni"]
 	path = vendor/github.com/containernetworking/cni
 	url = https://github.com/containernetworking/cni
+[submodule "vendor/google.golang.org/grpc"]
+	path = vendor/google.golang.org/grpc
+	url = https://github.com/grpc/grpc-go
+[submodule "vendor/github.com/docker/go-units"]
+	path = vendor/github.com/docker/go-units
+	url = https://github.com/docker/go-units
+[submodule "vendor/golang.org/x/net"]
+	path = vendor/golang.org/x/net
+	url = https://github.com/golang/net
+[submodule "vendor/k8s.io/api"]
+	path = vendor/k8s.io/api
+	url = https://github.com/kubernetes/api
+[submodule "vendor/k8s.io/apimachinery"]
+	path = vendor/k8s.io/apimachinery
+	url = https://github.com/kubernetes/apimachinery
+[submodule "vendor/github.com/PuerkitoBio/purell"]
+	path = vendor/github.com/PuerkitoBio/purell
+	url = https://github.com/PuerkitoBio/purell
+[submodule "vendor/github.com/PuerkitoBio/urlesc"]
+	path = vendor/github.com/PuerkitoBio/urlesc
+	url = https://github.com/PuerkitoBio/urlesc
+[submodule "vendor/github.com/emicklei/go-restful"]
+	path = vendor/github.com/emicklei/go-restful
+	url = https://github.com/emicklei/go-restful
+[submodule "vendor/github.com/go-openapi/jsonpointer"]
+	path = vendor/github.com/go-openapi/jsonpointer
+	url = https://github.com/go-openapi/jsonpointer
+[submodule "vendor/github.com/go-openapi/jsonreference"]
+	path = vendor/github.com/go-openapi/jsonreference
+	url = https://github.com/go-openapi/jsonreference
+[submodule "vendor/github.com/go-openapi/spec"]
+	path = vendor/github.com/go-openapi/spec
+	url = https://github.com/go-openapi/spec
+[submodule "vendor/github.com/go-openapi/swag"]
+	path = vendor/github.com/go-openapi/swag
+	url = https://github.com/go-openapi/swag
+[submodule "vendor/github.com/golang/glog"]
+	path = vendor/github.com/golang/glog
+	url = https://github.com/golang/glog
+[submodule "vendor/github.com/google/gofuzz"]
+	path = vendor/github.com/google/gofuzz
+	url = https://github.com/google/gofuzz
+[submodule "vendor/github.com/mailru/easyjson"]
+	path = vendor/github.com/mailru/easyjson
+	url = https://github.com/mailru/easyjson
+[submodule "vendor/k8s.io/kube-openapi"]
+	path = vendor/k8s.io/kube-openapi
+	url = https://github.com/kubernetes/kube-openapi
+[submodule "vendor/golang.org/x/text"]
+	path = vendor/golang.org/x/text
+	url = https://go.googlesource.com/text
+[submodule "vendor/gopkg.in/yaml.v2"]
+	path = vendor/gopkg.in/yaml.v2
+	url = https://gopkg.in/yaml.v2

CHANGELOG.md

@@ -1,3 +1,282 @@
+## Release 2.1.3
+
+This release fixes a race-condition in the IP reclaim code for weave-kube
+where two nodes could end up fighting over the same space and break
+connectivity #3190, #3192
+
+## Release 2.1.2
+
+This release fixes a couple of bugs discovered since the release of Weave Net 2.1.0
+
+##Bug fixes
+
+* Fix crash seen when starting 10-15 nodes simultaneously #3184,#3186
+* Fix NetworkPolicy blocking traffic if updates come out of order from Kubernetes #3177,#3181
+
+Thanks to the following contributors:
+@zignig
+
+[Full list of changes](https://github.com/weaveworks/weave/milestone/66?closed=1)
+
+
+## Release 2.1.1
+
+As 2.1.0, but fixing a couple of installation glitches. #3175,#3176
+
+## Release 2.1.0
+
+##New Features
+
+Improved Kubernetes Network Policy - Weave Net now supports the
+'v1' policies introduced in Kubernetes 1.7 as well as the 'beta'
+policies supported previously. See [Kubernetes 1.7 changelog](https://github.com/kubernetes/kubernetes/blob/master/CHANGELOG-1.7.md#network)
+for differences. To use old policies, `--use-legacy-netpol` argument
+should be passed to `weave-npc`. #3105,#3141,#3151,#3169
+
+Weave Net now reclaims IP addresses owned by Kubernetes nodes which
+have been deleted from the cluster - this avoids running out of IP
+addresses when many nodes are added and deleted over a long period.
+#2797,#3149,#3170,#3172
+
+##Other improvements
+
+* Export a Prometheus-style metric giving count of unreachable peers #3119
+* Update 'gopacket' library to reduce memory use by approx 15MB #3160
+* Replace bundling the 'docker' binary with our own code to avoid
+  security vulnerability alerts and save space #2957,#3110
+
+##Bug fixes
+
+* When `weave expose` is used, allow traffic into the Weave network -
+  up till version 1.12 Docker would do this for us, but in 1.13 they
+  stopped so now we do it. This change makes `weave expose` require
+  Weave Net to be running. #2758,#3122
+* Arm64 build now works on non-kubernetes installs #2832,#3110
+* TX offload was being disabled in 'awsvpc' mode, which slows down packet sending #3089
+* Removed spurious 'nil' in logs from CNI DEL operation #3143
+
+##Build and test
+
+* Images are now also built for the ppc64le platform #3129
+* Tweak build scripts to run on OSX as well as Linux #3135
+
+##External Contributors
+
+Thanks to the following contributors:
+@caarlos0
+@dtshepherd
+
+[Full list of changes](https://github.com/weaveworks/weave/milestone/59?closed=1)
+
+
+## Release 2.0.5
+
+Bug fixes
+
+* Fix /etc/hosts inside containers so the container's name resolves to
+  its Weave Net address #3136,#3138
+* New weave-kube config for Kubernetes 1.7 and 1.8 which resolves an
+  intermittent conflict with kube-proxy that could break Weave Net
+  until reboot #2998,#3134
+* Remove persistence file in `weave reset` when using Docker plugin V2 #3103,#3114
+
+Build and test
+
+* Some trivial changes to placate go-lint #3137
+
+[Full list of changes](https://github.com/weaveworks/weave/milestone/63?closed=1).
+
+
+## Release 2.0.4
+
+Bug fixes
+
+* weave-npc failed on Centos 7, due to older 'ipset' version in kernel #3099,#3100
+
+[Full list of changes](https://github.com/weaveworks/weave/milestone/62?closed=1).
+
+
+## Release 2.0.3
+
+Bug fixes
+
+* Weave-npc would crash on a policy with no 'from' part - regression introduced in 2.0.2 #3095,#3096,#3097
+
+[Full list of changes](https://github.com/weaveworks/weave/milestone/61?closed=1).
+
+
+## Release 2.0.2
+
+Bug fixes and minor improvements
+
+* Fix race condition in weave-npc which would intermittently block all traffic for a namespace  #3057,#3059
+* Ensure Fast Datapath works on machines that need to mount the kernel module dynamically #3080
+* Regression: weave-npc would block everything if `kubelet --hostname-override` was used #3049,#3051
+* Fix netfilter rules to block containers from accessing the Weave Net control endpoint #3093
+* If DNS server is off then disable proxy DNS registration, to avoid spurious errors #3054,#3088
+* Add comments to each iptables rule and ipset, to help when troubleshooting #3064
+* Remove code that checked for an outdated fallback address for Kubernetes api-server #3071
+* Add a label to the weavedb image so it can be filtered out by tools #3066
+* Fix various build and continuous-integration failures #3061
+* Print 'help' text faster in the weave script #3056
+* Add an option to create continuous integration hosts in different ways  #3060
+* Remove remnants of the pre-2.0 proxy and plugin from build and test #3035,#3036
+
+[Full list of changes](https://github.com/weaveworks/weave/milestone/60?closed=1).
+
+
+## Release 2.0.1
+
+Bug fixes and minor improvements
+
+* Fall back to slower data path (`sleeve`), rather than crashing, when the machine lacks VXLAN support (required for “fast data path”, `fastdp`)  #3043
+* Fix bug in processing of arguments when Docker has TLS enabled, rather than crashing with invalid peers list, e.g. `lookup --tlsverify: no such host` #3039
+* Add `kube-system` namespace back to `weave-kube`'s YAMLs, preventing omissions leading to errors like `error contacting APIServer: the server does not allow access to the requested resource` #3033,#3042
+* Fix release script to prevent ARM64 binaries to end up in AMD64 `net-plugin`, leading to `Error response from daemon: dial unix /run/docker/plugins/<id>/weave.sock: connect: no such file or directory` when installing `net-plugin` #3045
+* `weave reset` and `weave rmpeer` now only contact Weave Cloud when Weave Net is configured with a Weave Cloud token, preventing unnecessary requests and potentially confusing `401 Unauthorized` errors in Weave Net’s logs #3044
+
+[Full list of changes](https://github.com/weaveworks/weave/milestone/58?closed=1).
+
+
+## Release 2.0.0
+
+New Features
+
+Peer Discovery via Weave Cloud
+
+You can now get all your Weave Net peers to find each other via the
+Weave Cloud service, instead of maintaining a list of peers at
+startup. #2799,#2827
+
+See the [docs
+page](https://github.com/weaveworks/weave/blob/master/site/using-weave/weave-cloud.md)
+for more details
+
+New Docker Plugin
+
+Docker has a [new plugin
+system](https://docs.docker.com/engine/extend/) which improves the
+installation UX and solves some issues around startup.  This means
+Weave Net 2.0 can now run with Docker in "swarm mode" and supports the
+`docker service` command. #2396,#2397,#2651,#2727,#2805,#2816,#2905,
+#2906,#2929,#2932,#2945,#2950,#2956,#2963,#2964,#2966,#3019
+
+The previous Docker Plugin is still available and can be installed as before.
+
+All of Weave Net now runs in one container
+
+Previously we had three separate containers for routing, Docker API
+proxy and Docker plugin. Running everything in one simplifies start-up
+and removes the need to detect various error
+conditions. #1642,#2897,#2936,#2945,#2946,#2951,#2960
+
+The individual commands ‘weave launch-router’, ‘weave launch-plugin’,
+etc., have been removed. You can turn off the plugin and proxy with
+new command-line options. In keeping with Semantic Versioning, we have
+changed the major version number for this release.
+
+Other new features
+
+* Kubernetes configuration now comes from our “Launch Generator” that
+  allows different options to be selected via
+  URL. #2754,#2903,#3000,#3001
+* `weave-kube` now stores data about IP allocation in `/var/lib/weave`
+  on the host instead of in a Kubernetes volume.  This means that the
+  data will persist across pod deletion and re-creation, e.g. during
+  an upgrade of Weave Net, which makes restarts more
+  reliable. #2610,#2967
+* `weave-kube` turned on rolling updates, so careful manual handling
+  of updates is no longer required. #3024
+
+Bug fixes
+
+* Kubernetes Network Policies which allowed a specific set of pods to
+  connect would block all pods on other hosts. Revert the change in
+  v1.9.6 which ignored pods on other hosts #3025,#3028
+
+Features removed
+
+* `weave run` has been removed.  This was the original method provided
+  to start containers with Weave Net, but it always required care over
+  timing of start-up, and we now provide three alternative, better,
+  ways. You can replicate the effect by calling `docker run` then
+  `weave attach`.  Similarly `weave start` and `weave restart` were
+  removed. #2353,#2885
+* Everything deprecated more than one release ago has been removed, so
+  if you use it now you get an error rather than a warning. This
+  includes the ‘create-bridge’ command and older command-line
+  arguments, e.g. `--iprange` was replaced by `--ipalloc-range`
+  #2901,#2909,#2913,#2942,#2989,#2991
+
+Functions moved from shell-script to Go code.
+
+This enables more precise error-checking and runs a bit faster. It has
+also enabled us to shrink the size of images downloaded: `weave-kube`
+is 101MB compared to 163MB previously #2953,#2954,#2974
+
+Specific items that moved from shell-script to Go:
+* Setting up the `weave` bridge #1958,#2975,#2977,#2978
+* Container attachment #2947
+* Creation of the ’weave’ default plugin network #2920
+
+Minor improvements
+
+* You can now restart the Weave Net router without requiring the proxy to be enabled #2112
+* Plugin (legacy version) now respects `--ipalloc-default-subnet` option #2919
+* The `weave` script now detects and issues an error message if
+  `weave-kube` is running and you attempt to launch again from the script. #2709/#2966
+* It is now possible to choose the the MAC address of the `weave`
+  bridge using `--name`, in case your hosts have identical unique
+  IDs. #2900
+* Relaxed Kubernetes tolerations for Weave Net's daemonset in order to
+  match any node (previously, only taints directed at master). #3018
+* Kubernetes' `seLinuxOptions` configuration is now empty by default,
+  to reduce spurious failures on hosts not using seLinux. #3001
+* Improved reliability of namespace changes via `nsenter`. #2992
+* `weave ps` now fetches the list container IDs internally, rather
+  than calling out to `docker ps` #2814,#2898
+* at startup, actively remove dead containers’ Weave Net IP addresses from IPAM #3013
+* at startup, only check live containers to see if they have an
+  existing Weave Net IP address #2815,#2829
+* Weave Net CNI plugin now logs but does not raise an error if
+  anything goes wrong during network interface delete, to be more
+  compatible with Kubernetes 1.6. #2928
+* Stop running a shell in “privileged” mode when it’s only writing a file #2838
+* New internal REST endpoint to return all IP address mappings. #1350
+* Changed the wording where we do not log the password #2833
+* Fixed typo in plugin error messages #2894
+
+Build and test
+
+* Weave Net is now built with Go version 1.8, which has better code
+  generation and garbage collection #2914
+* During smoke-tests, use a webserver instead of just `ping` so we get
+  a more realistic test that the Weave network is working #2918
+* When installing dependencies for the build container, use a
+  keyserver port that's better for firewalls #2812
+* Kubernetes test script now scales up to more hosts, and works with
+  Kubernetes 1.6 #2837,#2853,#2923
+* Other minor build improvements and refactoring #2760,#2910
+
+[Full list of changes](https://github.com/weaveworks/weave/milestone/49?closed=1).
+
+
+## Release 1.9.8
+
+Bug fixes and minor improvements
+
+* Fix weave-npc blocking NodePort and any other non-local access #3011,#3014
+* Fix bug where IPAM would duplicate a fixed IP address assigned via Docker plugin #3003,#3010
+
+[Full list of changes](https://github.com/weaveworks/weave/milestone/56?closed=1).
+
+
+## Release 1.9.7
+
+This is identical to 1.9.6 with one additional bug-fix:
+
+* weave-npc would block everything if `kubelet --hostname-override` was used #2995,#2996
+
 ## Release 1.9.6
 
 Bug fixes and minor improvements

DEPENDENCIES

@@ -1,3 +1,3 @@
-DOCKER_VERSION=1.11.2
-KUBERNETES_VERSION=1.5.2
-KUBERNETES_CNI_VERSION=0.3.0.1
+DOCKER_VERSION=17.06
+KUBERNETES_VERSION=1.7.8
+KUBERNETES_CNI_VERSION=0.5.1

LICENSE

@@ -176,7 +176,7 @@
 
    END OF TERMS AND CONDITIONS
 
-   Copyright 2014-2016 Weaveworks Ltd.
+   Copyright 2014-2017 Weaveworks Ltd.
 
    Licensed under the Apache License, Version 2.0 (the "License");
    you may not use this file except in compliance with the License.

Makefile

@@ -1,5 +1,5 @@
 .DEFAULT: all
-.PHONY: all exes testrunner update tests lint publish-one-arch $(PUBLISH) clean clean-bin prerequisites build run-smoketests
+.PHONY: all exes testrunner update tests lint publish-one-arch $(PUBLISH) clean clean-bin clean-work-dir prerequisites build run-smoketests
 
 # If you can use docker without being root, you can do "make SUDO="
 SUDO=$(shell docker info >/dev/null 2>&1 || echo "sudo -E")
@@ -17,15 +17,15 @@ QEMU_VERSION=v2.7.0
 
 # A list of all supported architectures here. Should be named as Go is naming platforms
 # All supported architectures must have an "ifeq" block below that customizes the parameters
-ALL_ARCHITECTURES=amd64 arm arm64
-ML_PLATFORMS=linux/amd64,linux/arm,linux/arm64
+ALL_ARCHITECTURES=amd64 arm arm64 ppc64le
+ML_PLATFORMS=linux/amd64,linux/arm,linux/arm64,linux/ppc64le
 
 ifeq ($(ARCH),amd64)
 # The architecture to use when downloading the docker binary
 	WEAVEEXEC_DOCKER_ARCH?=x86_64
 
 # The name of the alpine baseimage to use as the base for weave images
-	ALPINE_BASEIMAGE?=alpine:3.4
+	ALPINE_BASEIMAGE?=alpine:3.6
 
 # The extension for the made images
 # Specifying none means for example weaveworks/weave:latest
@@ -42,7 +42,7 @@ ifeq ($(ARCH),arm)
 	WEAVEEXEC_DOCKER_ARCH?=armel
 
 # Using the (semi-)official alpine image
-	ALPINE_BASEIMAGE?=armhf/alpine:3.4
+	ALPINE_BASEIMAGE?=arm32v6/alpine:3.6
 
 # arm images have the -arm suffix, for instance weaveworks/weave-arm:latest
 	ARCH_EXT?=-arm
@@ -62,7 +62,7 @@ ifeq ($(ARCH),arm64)
 	WEAVEEXEC_DOCKER_ARCH?=armel
 
 # Using the (semi-)official alpine image
-	ALPINE_BASEIMAGE?=aarch64/alpine:3.5
+	ALPINE_BASEIMAGE?=arm64v8/alpine:3.6
 
 # arm64 images have the -arm64 suffix, for instance weaveworks/weave-arm64:latest
 	ARCH_EXT?=-arm64
@@ -77,15 +77,36 @@ ifeq ($(ARCH),arm64)
 # Tell the gcc linker to search for libpcap here
 	CGO_LDFLAGS="-L/usr/local/lib/$(CC)"
 endif
+ifeq ($(ARCH),ppc64le)
+# The architecture to use when downloading the docker binary
+	WEAVEEXEC_DOCKER_ARCH?=ppc64le
+
+# Using the (semi-)official alpine image
+	ALPINE_BASEIMAGE?=ppc64le/alpine:3.6
+
+# ppc64le images have the -ppc64le suffix, for instance weaveworks/weave-ppc64le:latest
+	ARCH_EXT?=-ppc64le
+
+# The name of the gcc binary
+	CC=powerpc64le-linux-gnu-gcc
+
+# The architecture name to use when downloading a prebuilt QEMU binary
+	QEMUARCH=ppc64le
+
+# In the weaveworks/build image; libpcap libraries for ppc64le are placed here
+# Tell the gcc linker to search for libpcap here
+	CGO_LDFLAGS="-L/usr/local/lib/$(CC)"
+endif
 
 # The name of the user that this Makefile should produce image artifacts for. Can/should be overridden
 DOCKERHUB_USER?=weaveworks
 # The default version that's chosen when pushing the images. Can/should be overridden
 WEAVE_VERSION?=git-$(shell git rev-parse --short=12 HEAD)
+# Docker Store does not allow the "latest" tag.
+NET_PLUGIN_LATEST=latest_release
 
 # Paths to all relevant binaries that should be compiled
 WEAVER_EXE=prog/weaver/weaver
-WEAVEPROXY_EXE=prog/weaveproxy/weaveproxy
 SIGPROXY_EXE=prog/sigproxy/sigproxy
 KUBEPEERS_EXE=prog/kube-peers/kube-peers
 WEAVENPC_EXE=prog/weave-npc/weave-npc
@@ -93,45 +114,47 @@ WEAVEWAIT_EXE=prog/weavewait/weavewait
 WEAVEWAIT_NOOP_EXE=prog/weavewait/weavewait_noop
 WEAVEWAIT_NOMCAST_EXE=prog/weavewait/weavewait_nomcast
 WEAVEUTIL_EXE=prog/weaveutil/weaveutil
-PLUGIN_EXE=prog/plugin/plugin
 RUNNER_EXE=tools/runner/runner
 MANIFEST_TOOL_DIR=vendor/github.com/estesp/manifest-tool
 MANIFEST_TOOL_EXE=$(MANIFEST_TOOL_DIR)/manifest-tool
 TEST_TLS_EXE=test/tls/tls
+NETWORKTESTER_EXE=test/images/network-tester/webserver
 
 # All binaries together in a list
-EXES=$(WEAVER_EXE) $(SIGPROXY_EXE) $(KUBEPEERS_EXE) $(WEAVENPC_EXE) $(WEAVEPROXY_EXE) $(WEAVEWAIT_EXE) $(WEAVEWAIT_NOOP_EXE) $(WEAVEWAIT_NOMCAST_EXE) $(WEAVEUTIL_EXE) $(PLUGIN_EXE) $(RUNNER_EXE) $(TEST_TLS_EXE) $(MANIFEST_TOOL_EXE)
+EXES=$(WEAVER_EXE) $(SIGPROXY_EXE) $(KUBEPEERS_EXE) $(WEAVENPC_EXE) $(WEAVEWAIT_EXE) $(WEAVEWAIT_NOOP_EXE) $(WEAVEWAIT_NOMCAST_EXE) $(WEAVEUTIL_EXE) $(RUNNER_EXE) $(TEST_TLS_EXE) $(MANIFEST_TOOL_EXE) $(NETWORKTESTER_EXE)
 
 # These stamp files are used to mark the current state of the build; whether an image has been built or not
 BUILD_UPTODATE=.build.uptodate
 WEAVER_UPTODATE=.weaver$(ARCH_EXT).uptodate
 WEAVEEXEC_UPTODATE=.weaveexec$(ARCH_EXT).uptodate
-PLUGIN_UPTODATE=.plugin$(ARCH_EXT).uptodate
+PLUGIN_UPTODATE=.net-plugin$(ARCH_EXT).uptodate
 WEAVEKUBE_UPTODATE=.weavekube$(ARCH_EXT).uptodate
 WEAVENPC_UPTODATE=.weavenpc$(ARCH_EXT).uptodate
 WEAVEDB_UPTODATE=.weavedb.uptodate
+NETWORKTESTER_UPTODATE=.networktester$(ARCH_EXT).uptodate
 
-IMAGES_UPTODATE=$(WEAVER_UPTODATE) $(WEAVEEXEC_UPTODATE) $(PLUGIN_UPTODATE) $(WEAVEKUBE_UPTODATE) $(WEAVENPC_UPTODATE)
+IMAGES_UPTODATE=$(WEAVER_UPTODATE) $(WEAVEEXEC_UPTODATE) $(WEAVEKUBE_UPTODATE) $(WEAVENPC_UPTODATE) $(NETWORKTESTER_UPTODATE)
 
 # The names of the images. Note that the images for other architectures than amd64 have a suffix in the image name.
 WEAVER_IMAGE=$(DOCKERHUB_USER)/weave$(ARCH_EXT)
 WEAVEEXEC_IMAGE=$(DOCKERHUB_USER)/weaveexec$(ARCH_EXT)
-PLUGIN_IMAGE=$(DOCKERHUB_USER)/plugin$(ARCH_EXT)
 WEAVEKUBE_IMAGE=$(DOCKERHUB_USER)/weave-kube$(ARCH_EXT)
 WEAVENPC_IMAGE=$(DOCKERHUB_USER)/weave-npc$(ARCH_EXT)
 BUILD_IMAGE=weaveworks/weavebuild
 WEAVEDB_IMAGE=$(DOCKERHUB_USER)/weavedb
+PLUGIN_IMAGE=$(DOCKERHUB_USER)/net-plugin
+NETWORKTESTER_IMAGE=$(DOCKERHUB_USER)/network-tester
 
-IMAGES=$(WEAVER_IMAGE) $(WEAVEEXEC_IMAGE) $(PLUGIN_IMAGE) $(WEAVEKUBE_IMAGE) $(WEAVENPC_IMAGE) $(WEAVEDB_IMAGE)
+IMAGES=$(WEAVER_IMAGE) $(WEAVEEXEC_IMAGE) $(WEAVEKUBE_IMAGE) $(WEAVENPC_IMAGE) $(WEAVEDB_IMAGE) $(NETWORKTESTER_IMAGE)
 
-PUBLISH=publish_weave publish_weaveexec publish_plugin publish_weave-kube publish_weave-npc
-PUSH_ML=push_ml_weave push_ml_weaveexec push_ml_plugin push_ml_weave-kube push_ml_weave-npc
+PLUGIN_WORK_DIR="prog/net-plugin/rootfs"
+PLUGIN_BUILD_IMG="plugin-builder"
+
+PUBLISH=publish_weave publish_weaveexec publish_weave-kube publish_weave-npc
+PUSH_ML=push_ml_weave push_ml_weaveexec push_ml_weave-kube push_ml_weave-npc
 
 WEAVE_EXPORT=weave$(ARCH_EXT).tar.gz
 
-DOCKER_VERSION=1.10.3
-DOCKER_DISTRIB=prog/weaveexec/docker-$(DOCKER_VERSION).tgz
-DOCKER_DISTRIB_URL=https://get.docker.com/builds/Linux/$(WEAVEEXEC_DOCKER_ARCH)/docker-$(DOCKER_VERSION).tgz
 NETGO_CHECK=@strings $@ | grep cgo_stub\\\.go >/dev/null || { \
 	rm $@; \
 	echo "\nYour go standard library was built without the 'netgo' build tag."; \
@@ -150,20 +173,21 @@ PACKAGE_BASE=$(shell go list -e ./)
 all: $(WEAVE_EXPORT)
 testrunner: $(RUNNER_EXE) $(TEST_TLS_EXE)
 
-$(WEAVER_EXE) $(WEAVEPROXY_EXE) $(WEAVEUTIL_EXE): common/*.go common/*/*.go net/*.go net/*/*.go
+$(WEAVER_EXE) $(WEAVEUTIL_EXE): common/*.go common/*/*.go net/*.go net/*/*.go
 $(WEAVER_EXE): router/*.go ipam/*.go ipam/*/*.go db/*.go nameserver/*.go prog/weaver/*.go
-$(WEAVEPROXY_EXE): proxy/*.go prog/weaveproxy/*.go
+$(WEAVER_EXE): api/*.go plugin/*.go plugin/*/*
+$(WEAVER_EXE):  proxy/*.go
 $(WEAVEUTIL_EXE): prog/weaveutil/*.go net/*.go plugin/net/*.go plugin/ipam/*.go db/*.go
 $(SIGPROXY_EXE): prog/sigproxy/*.go
 $(KUBEPEERS_EXE): prog/kube-peers/*.go
 $(WEAVENPC_EXE): prog/weave-npc/*.go npc/*.go npc/*/*.go
-$(PLUGIN_EXE): prog/plugin/*.go plugin/*/*.go api/*.go common/*.go common/docker/*.go net/*.go
 $(TEST_TLS_EXE): test/tls/*.go
 $(RUNNER_EXE): tools/runner/*.go
 $(MANIFEST_TOOL_EXE): $(MANIFEST_TOOL_DIR)/*.go
 $(WEAVEWAIT_NOOP_EXE): prog/weavewait/*.go
 $(WEAVEWAIT_EXE): prog/weavewait/*.go net/*.go
 $(WEAVEWAIT_NOMCAST_EXE): prog/weavewait/*.go net/*.go
+$(NETWORKTESTER_EXE): test/images/network-tester/*.go
 tests: tools/.git
 lint: tools/.git
 
@@ -181,12 +205,13 @@ exes $(EXES) tests lint: $(BUILD_UPTODATE)
 		-v $(shell pwd)/.pkg:/go/pkg \
 		-e GOARCH=$(ARCH) -e CGO_ENABLED=1 -e GOOS=linux -e CIRCLECI -e CIRCLE_BUILD_NUM -e CIRCLE_NODE_TOTAL -e CIRCLE_NODE_INDEX -e COVERDIR -e SLOW -e DEBUG \
 		$(BUILD_IMAGE) COVERAGE=$(COVERAGE) WEAVE_VERSION=$(WEAVE_VERSION) CC=$(CC) QEMUARCH=$(QEMUARCH) CGO_LDFLAGS=$(CGO_LDFLAGS) $@
+	touch $@
 
 else
 
 exes: $(EXES)
 
-$(WEAVER_EXE) $(WEAVEPROXY_EXE) $(PLUGIN_EXE):
+$(WEAVER_EXE):
 ifeq ($(COVERAGE),true)
 	$(eval COVERAGE_MODULES := $(shell (go list ./$(@D); go list -f '{{join .Deps "\n"}}' ./$(@D) | grep "^$(PACKAGE_BASE)/") | grep -v "^$(PACKAGE_BASE)/vendor/" | paste -s -d,))
 	go test -c -o ./$@ $(BUILD_FLAGS) -v -covermode=atomic -coverpkg $(COVERAGE_MODULES) ./$(@D)/
@@ -195,7 +220,7 @@ else
 endif
 	$(NETGO_CHECK)
 
-$(WEAVEUTIL_EXE) $(KUBEPEERS_EXE) $(WEAVENPC_EXE):
+$(WEAVEUTIL_EXE) $(KUBEPEERS_EXE) $(WEAVENPC_EXE) $(NETWORKTESTER_EXE):
 	go build $(BUILD_FLAGS) -o $@ ./$(@D)
 	$(NETGO_CHECK)
 
@@ -236,35 +261,45 @@ $(BUILD_UPTODATE): build/*
 	sed -e "s|DOCKERHUB_USER|$(DOCKERHUB_USER)|g;s|ARCH_EXT|$(ARCH_EXT)|g;s|ALPINE_BASEIMAGE|$(ALPINE_BASEIMAGE)|g;s|QEMUARCH|$(QEMUARCH)|g" $^ > $@
 ifeq ($(ARCH),amd64)
 # When building "normally" for amd64, remove the whole line, it has no part in the amd64 image
-	sed -i "/CROSS_BUILD_/d" $@
+	sed -i.bak "/CROSS_BUILD_/d" $@
 else
 # When cross-building, only the placeholder "CROSS_BUILD_" should be removed
-# Register /usr/bin/qemu-ARCH-static as the handler for ARM binaries in the kernel
+# Register /usr/bin/qemu-ARCH-static as the handler for ARM and ppc64le binaries in the kernel
 	curl -sSL https://github.com/multiarch/qemu-user-static/releases/download/$(QEMU_VERSION)/x86_64_qemu-$(QEMUARCH)-static.tar.gz | tar -xz -C $(shell dirname $@)
 	cd $(shell dirname $@) && sha256sum -c $(shell pwd)/build/shasums/qemu-$(QEMUARCH)-static.sha256sum
-	sed -i "s/CROSS_BUILD_//g" $@
+	sed -i.bak "s/CROSS_BUILD_//g" $@
 endif
 
 # The targets below builds the weave images
-$(WEAVER_UPTODATE): prog/weaver/Dockerfile.$(DOCKERHUB_USER) $(WEAVER_EXE) $(WEAVEEXEC_UPTODATE)
+$(WEAVER_UPTODATE): prog/weaver/Dockerfile.$(DOCKERHUB_USER) $(WEAVER_EXE) weave $(WEAVEUTIL_EXE)
+	cp $(WEAVEUTIL_EXE) prog/weaver/weaveutil
+	cp weave prog/weaver/weave
 	$(SUDO) DOCKER_HOST=$(DOCKER_HOST) docker build -f prog/weaver/Dockerfile.$(DOCKERHUB_USER) -t $(WEAVER_IMAGE) prog/weaver
 	touch $@
 
-$(WEAVEEXEC_UPTODATE): prog/weaveexec/Dockerfile.$(DOCKERHUB_USER) prog/weaveexec/symlink $(DOCKER_DISTRIB) weave $(SIGPROXY_EXE) $(WEAVEPROXY_EXE) $(WEAVEWAIT_EXE) $(WEAVEWAIT_NOOP_EXE) $(WEAVEWAIT_NOMCAST_EXE) $(WEAVEUTIL_EXE)
-	cp weave prog/weaveexec/weave
+$(WEAVEEXEC_UPTODATE): prog/weaveexec/Dockerfile.$(DOCKERHUB_USER) prog/weaveexec/symlink $(SIGPROXY_EXE) $(WEAVEWAIT_EXE) $(WEAVEWAIT_NOOP_EXE) $(WEAVEWAIT_NOMCAST_EXE) $(WEAVER_UPTODATE)
 	cp $(SIGPROXY_EXE) prog/weaveexec/sigproxy
-	cp $(WEAVEPROXY_EXE) prog/weaveexec/weaveproxy
 	cp $(WEAVEWAIT_EXE) prog/weaveexec/weavewait
 	cp $(WEAVEWAIT_NOOP_EXE) prog/weaveexec/weavewait_noop
 	cp $(WEAVEWAIT_NOMCAST_EXE) prog/weaveexec/weavewait_nomcast
-	cp $(WEAVEUTIL_EXE) prog/weaveexec/weaveutil
-	tar -xf $(DOCKER_DISTRIB) usr/local/bin/docker -O > prog/weaveexec/docker
-	chmod +x prog/weaveexec/docker
 	$(SUDO) DOCKER_HOST=$(DOCKER_HOST) docker build -f prog/weaveexec/Dockerfile.$(DOCKERHUB_USER) -t $(WEAVEEXEC_IMAGE) prog/weaveexec
 	touch $@
 
-$(PLUGIN_UPTODATE): prog/plugin/Dockerfile.$(DOCKERHUB_USER) $(PLUGIN_EXE) $(WEAVER_UPTODATE)
-	$(SUDO) docker build -f prog/plugin/Dockerfile.$(DOCKERHUB_USER) -t $(PLUGIN_IMAGE) prog/plugin
+# Builds Docker plugin.
+$(PLUGIN_UPTODATE): prog/net-plugin/launch.sh prog/net-plugin/config.json $(WEAVER_UPTODATE)
+	-$(SUDO) docker rm -f $(PLUGIN_BUILD_IMG) 2>/dev/null
+	$(SUDO) docker create --name=$(PLUGIN_BUILD_IMG) $(WEAVER_IMAGE) true
+	rm -rf $(PLUGIN_WORK_DIR)
+	mkdir $(PLUGIN_WORK_DIR)
+	$(SUDO) docker export $(PLUGIN_BUILD_IMG) | tar -x -C $(PLUGIN_WORK_DIR)
+	$(SUDO) docker rm -f $(PLUGIN_BUILD_IMG)
+	cp prog/net-plugin/launch.sh $(PLUGIN_WORK_DIR)/home/weave/launch.sh
+	-$(SUDO) docker plugin disable $(PLUGIN_IMAGE):$(WEAVE_VERSION) 2>/dev/null
+	-$(SUDO) docker plugin rm $(PLUGIN_IMAGE):$(WEAVE_VERSION) 2>/dev/null
+	$(SUDO) docker plugin create $(PLUGIN_IMAGE):$(WEAVE_VERSION) prog/net-plugin
+	-$(SUDO) docker plugin disable $(PLUGIN_IMAGE):$(NET_PLUGIN_LATEST) 2>/dev/null
+	-$(SUDO) docker plugin rm $(PLUGIN_IMAGE):$(NET_PLUGIN_LATEST) 2>/dev/null
+	$(SUDO) docker plugin create $(PLUGIN_IMAGE):$(NET_PLUGIN_LATEST) prog/net-plugin
 	touch $@
 
 $(WEAVEKUBE_UPTODATE): prog/weave-kube/Dockerfile.$(DOCKERHUB_USER) prog/weave-kube/launch.sh $(KUBEPEERS_EXE) $(WEAVER_UPTODATE)
@@ -280,18 +315,26 @@ $(WEAVEDB_UPTODATE): prog/weavedb/Dockerfile
 	$(SUDO) docker build -t $(WEAVEDB_IMAGE) prog/weavedb
 	touch $@
 
+$(NETWORKTESTER_UPTODATE): test/images/network-tester/Dockerfile $(NETWORKTESTER_EXE)
+	$(SUDO) docker build -t $(NETWORKTESTER_IMAGE) test/images/network-tester
+	touch $@
+
 $(WEAVE_EXPORT): $(IMAGES_UPTODATE) $(WEAVEDB_UPTODATE)
 	$(SUDO) DOCKER_HOST=$(DOCKER_HOST) docker save $(addsuffix :latest,$(IMAGES)) | gzip > $@
 
-$(DOCKER_DISTRIB):
-	curl -o $(DOCKER_DISTRIB) $(DOCKER_DISTRIB_URL)
-	cd $(shell dirname $@) && sha256sum -c $(shell pwd)/build/shasums/docker-tgz-$(WEAVEEXEC_DOCKER_ARCH).sha256sum
-
 tools/.git $(MANIFEST_TOOL_DIR)/.git:
 	git submodule update --init
 
 # CODE FOR PUBLISHING THE IMAGES
 
+# Push plugin
+plugin_publish: $(PLUGIN_UPTODATE)
+	$(SUDO) DOCKER_HOST=$(DOCKER_HOST) docker plugin push $(PLUGIN_IMAGE):$(WEAVE_VERSION)
+# "$(NET_PLUGIN_LATEST)" means "stable release" here, so only push that if explicitly told to
+ifeq ($(UPDATE_LATEST),true)
+	$(SUDO) DOCKER_HOST=$(DOCKER_HOST) docker plugin push $(PLUGIN_IMAGE):$(NET_PLUGIN_LATEST)
+endif
+
 # This target first runs "make publish" for each architecture
 # Then it pushes the manifest lists
 publish: $(addprefix sub-publish-,$(ALL_ARCHITECTURES)) $(WEAVEDB_UPTODATE)
@@ -311,7 +354,7 @@ publish-one-arch: $(PUBLISH)
 $(PUBLISH): publish_%: $(IMAGES_UPTODATE)
 # Tag :latest with the real version
 	$(SUDO) DOCKER_HOST=$(DOCKER_HOST) docker tag  $(DOCKERHUB_USER)/$*$(ARCH_EXT) $(DOCKERHUB_USER)/$*$(ARCH_EXT):$(WEAVE_VERSION)
-# Push the image with the arch suffix for arm and arm64, and without suffix for amd64
+# Push the image with the arch suffix for arm, arm64 and ppc64le, and without suffix for amd64
 	$(MAKE) DOCKER_HOST=$(DOCKER_HOST) DOCKERHUB_USER=$(DOCKERHUB_USER) WEAVE_VERSION=$(WEAVE_VERSION) UPDATE_LATEST=$(UPDATE_LATEST) push_$*$(ARCH_EXT)
 
 ifeq ($(ARCH),amd64)
@@ -344,13 +387,16 @@ ifneq ($(UPDATE_LATEST),false)
 	$(MANIFEST_TOOL_EXE) push from-args --platforms $(ML_PLATFORMS) --template $(DOCKERHUB_USER)/$*-ARCH:latest --target $(DOCKERHUB_USER)/$*:latest
 endif
 
+clean-work-dir:
+	rm -rf $(PLUGIN_WORK_DIR)
+
 clean-bin:
 	-$(SUDO) DOCKER_HOST=$(DOCKER_HOST) docker rmi $(IMAGES)
 	find prog -type f -name "Dockerfile.*" -not -name "Dockerfile.template" -print | xargs rm -f
 	find prog -type f -name "*qemu-*" -print | xargs rm -f
-	rm -rf $(EXES) $(IMAGES_UPTODATE) $(WEAVEDB_UPTODATE) weave*.tar.gz $(DOCKER_DISTRIB) prog/weaveexec/docker .pkg
+	rm -rf $(EXES) $(IMAGES_UPTODATE) $(WEAVEDB_UPTODATE) weave*.tar.gz .pkg
 
-clean: clean-bin
+clean: clean-bin clean-work-dir
 	-$(SUDO) DOCKER_HOST=$(DOCKER_HOST) docker rmi $(BUILD_IMAGE)
 	rm -rf test/tls/*.pem test/coverage.* test/coverage $(BUILD_UPTODATE) $(MANIFEST_TOOL_EXE)
 

README.md

@@ -3,7 +3,6 @@
 [![Integration Tests](https://circleci.com/gh/weaveworks/weave/tree/master.svg?style=shield)](https://circleci.com/gh/weaveworks/weave)
 [![Coverage Status](https://coveralls.io/repos/weaveworks/weave/badge.svg)](https://coveralls.io/r/weaveworks/weave)
 [![Go Report Card](https://goreportcard.com/badge/github.com/weaveworks/weave)](https://goreportcard.com/report/github.com/weaveworks/weave)
-[![Slack Status](https://slack.weave.works/badge.svg)](https://slack.weave.works)
 [![Docker Pulls](https://img.shields.io/docker/pulls/weaveworks/weave.svg?maxAge=604800)](https://hub.docker.com/r/weaveworks/weave/)
 
 # About Weaveworks
@@ -11,11 +10,13 @@
 [Weaveworks](https://www.weave.works) is the company that delivers the most productive way for developers to connect, observe and control
 Docker containers.
 
-This repository contains [Weave Net](https://www.weave.works/products/weave-net/), the first product developed by Weaveworks, and with over 8 million downloads to date, it enables you to get started with Docker clusters and portable apps in a fraction of the time compared with other solutions.
+This repository contains [Weave Net](https://www.weave.works/products/weave-net/), the first product developed by Weaveworks, and with over 8 million downloads to date, enables you to get started with Docker clusters and portable apps in a fraction of the time compared with other solutions.
 
-Weaveworks' product [Weave Cloud](https://www.weave.works/solution/cloud/) takes the heavy lifting out of building, integrating and operating open source components, so you don’t have to manage databases, storage and availability.
+## Weave Cloud: 
 
-Weave Cloud is built using these Open Source projects: [Weave Scope](https://www.weave.works/products/weave-scope/), a powerful container monitoring tool that automatically maps Docker containers and their interactions, [Weave Cortex](https://github.com/weaveworks/cortex), a horizontally-scalable version of Prometheus, and [Weave Flux](https://www.weave.works/products/weave-flux/), a continuous deployment tool that works with Kubernetes.
+[Weave Cloud](https://www.weave.works/product/cloud/) is software-as-a-service that simplifies deployment, monitoring and management for containers and microservices. It extends and complements popular orchestrators, and enables developers and DevOps to realize faster deployments, insightful monitoring, visualization and networking.
+
+Weave Cloud is built using these Open Source projects: [Weave Scope](https://www.weave.works/products/weave-scope/), a powerful container visualization tool that automatically maps Docker containers and their interactions, [Weave Cortex](https://github.com/weaveworks/cortex), a horizontally-scalable version of Prometheus, and [Weave Flux](https://www.weave.works/products/weave-flux/), a continuous deployment tool that works with Kubernetes.
 
 # Weave Net
 
@@ -27,7 +28,7 @@ Services provided by application containers on the Weave network can be exposed
 
 If you have any questions about, feedback for or a problem with Weave Net:
 
-- Read [the Weave Net docs](https://www.weave.works/docs/net/latest/introducing-weave/).
+- Read [the Weave Net docs](https://www.weave.works/docs/net/latest/overview).
 - Invite yourself to the <a href="https://weaveworks.github.io/community-slack/" target="_blank"> #weave-community </a> slack channel.
 - Ask a question on the <a href="https://weave-community.slack.com/messages/general/"> #weave-community</a> slack channel.
 - Join the <a href="https://www.meetup.com/pro/Weave/"> Weave User Group </a> and get invited to online talks, hands-on training and meetups in your area.
@@ -36,7 +37,7 @@ If you have any questions about, feedback for or a problem with Weave Net:
 
 Your feedback is always welcome!
 
-##Further Information
+## Further Information
 
-* [Installing Weave Net](https://www.weave.works/docs/net/latest/installing-weave/)
-* [Weave Net Product Overview](https://www.weave.works/products/weave-net/)
+* [Installing Weave Net](https://www.weave.works/docs/net/latest/install/installing-weave)
+* [Weave Net Product Overview](https://www.weave.works/products/weave-net)

api/ipam.go

@@ -3,6 +3,7 @@ package api
 import (
 	"fmt"
 	"net"
+	"net/url"
 )
 
 // Special token used in place of a container identifier when:
@@ -14,21 +15,29 @@ import (
 // which should be kept in mind if the entry needs to be removed at any point.
 const NoContainerID string = "_"
 
-func (client *Client) ipamOp(ID string, op string) (*net.IPNet, error) {
-	ip, err := client.httpVerb(op, fmt.Sprintf("/ip/%s", ID), nil)
+func (client *Client) ipamOp(ID string, op string, values url.Values) (*net.IPNet, error) {
+	ip, err := client.httpVerb(op, fmt.Sprintf("/ip/%s", ID), values)
 	if err != nil {
 		return nil, err
 	}
 	return parseIP(ip)
 }
 
-// returns an IP for the ID given, allocating a fresh one if necessary
-func (client *Client) AllocateIP(ID string) (*net.IPNet, error) {
-	return client.ipamOp(ID, "POST")
+func ipamValues(checkAlive bool) url.Values {
+	values := make(url.Values)
+	if checkAlive {
+		values.Set("check-alive", "true")
+	}
+	return values
 }
 
-func (client *Client) AllocateIPInSubnet(ID string, subnet *net.IPNet) (*net.IPNet, error) {
-	ip, err := client.httpVerb("POST", fmt.Sprintf("/ip/%s/%s", ID, subnet), nil)
+// returns an IP for the ID given, allocating a fresh one if necessary
+func (client *Client) AllocateIP(ID string, checkAlive bool) (*net.IPNet, error) {
+	return client.ipamOp(ID, "POST", ipamValues(checkAlive))
+}
+
+func (client *Client) AllocateIPInSubnet(ID string, subnet *net.IPNet, checkAlive bool) (*net.IPNet, error) {
+	ip, err := client.httpVerb("POST", fmt.Sprintf("/ip/%s/%s", ID, subnet), ipamValues(checkAlive))
 	if err != nil {
 		return nil, err
 	}
@@ -38,12 +47,12 @@ func (client *Client) AllocateIPInSubnet(ID string, subnet *net.IPNet) (*net.IPN
 // returns an IP for the ID given, or nil if one has not been
 // allocated
 func (client *Client) LookupIP(ID string) (*net.IPNet, error) {
-	return client.ipamOp(ID, "GET")
+	return client.ipamOp(ID, "GET", nil)
 }
 
 // Claim a specific IP on behalf of the ID
-func (client *Client) ClaimIP(ID string, cidr *net.IPNet) error {
-	_, err := client.httpVerb("PUT", fmt.Sprintf("/ip/%s/%s", ID, cidr), nil)
+func (client *Client) ClaimIP(ID string, cidr *net.IPNet, checkAlive bool) error {
+	_, err := client.httpVerb("PUT", fmt.Sprintf("/ip/%s/%s", ID, cidr), ipamValues(checkAlive))
 	return err
 }
 
@@ -53,6 +62,12 @@ func (client *Client) ReleaseIPsFor(ID string) error {
 	return err
 }
 
+// release all IP space owned by a peer
+func (client *Client) RmPeer(peerName string) (string, error) {
+	result, err := client.httpVerb("DELETE", fmt.Sprintf("/peer/%s", peerName), nil)
+	return result, err
+}
+
 func (client *Client) DefaultSubnet() (*net.IPNet, error) {
 	cidr, err := client.httpVerb("GET", fmt.Sprintf("/ipinfo/defaultsubnet"), nil)
 	if err != nil {

api/router.go

@@ -0,0 +1,12 @@
+package api
+
+import (
+	"fmt"
+	"net"
+)
+
+// Expose calls the router to assign the given IP addr to the weave bridge.
+func (client *Client) Expose(ipAddr *net.IPNet) error {
+	_, err := client.httpVerb("POST", fmt.Sprintf("/expose/%s", ipAddr), nil)
+	return err
+}

api/weaveapi.go

@@ -8,6 +8,7 @@ import (
 	"net/http"
 	"net/url"
 	"strings"
+	"time"
 )
 
 const (
@@ -76,6 +77,22 @@ func (client *Client) Connect(remote string) error {
 	return err
 }
 
+// IsReady returns true if the API server is up and running
+// (note it returns StatusServiceUnavailable until all parts are ready,
+// but callers of this function are expected to be ok with a partial service)
+func (client *Client) IsReady() bool {
+	resp, err := http.Get(client.baseURL + "/status")
+	return err == nil &&
+		(resp.StatusCode == http.StatusOK || resp.StatusCode == http.StatusServiceUnavailable)
+}
+
+// WaitAPIServer waits until the API server is ready to serve.
+func (client *Client) WaitAPIServer(n int) {
+	for i := 0; i < n && !client.IsReady(); i++ {
+		time.Sleep(2 * time.Second)
+	}
+}
+
 type Logger interface {
 	Infof(string, ...interface{})
 	Debugf(string, ...interface{})

bin/multiweave

@@ -41,7 +41,7 @@ case "$1" in
         for i in $(seq $START $FINISH); do
             IP=$SUBNET.$i
             sudo ip addr add $IP/24 dev multiweave
-            weave $i launch-router --datapath '' --iface '' --no-dns --host $IP \
+            weave $i launch --datapath '' --iface '' --no-dns --host $IP \
                 --name $(random_mac) --nickname weave$i "$@"
         done
         echo $(date -R) connecting
@@ -57,7 +57,7 @@ case "$1" in
         for i in $(seq $FINISH -1 $START); do
             IP=$SUBNET.$i
             echo Stopping weave$i
-            weave $i stop-router
+            weave $i stop
             sudo ip addr del $IP/24 dev multiweave || true
         done
         [ -z "$(ip -4 -o addr show multiweave)" ] && sudo ip link del multiweave

bin/provision_test_vms.sh

@@ -17,7 +17,9 @@ function install_terraform() {
 
 function install_ansible() {
     sudo apt-get update || true
-    sudo apt-get install -qq -y python-pip python-dev libffi-dev libssl-dev && pip install --user -U cffi && pip install --user ansible
+    sudo apt-get install -qq -y python-pip python-dev libffi-dev libssl-dev \
+        && pip install --user -U setuptools cffi \
+        && pip install --user ansible
     export PATH="$PATH:$HOME/.local/bin"
 }
 

bin/release

@@ -88,9 +88,10 @@ build() {
 
     ## Inject the version numbers and build the distributables
     ## (library versions?)
-    sed -i "/SCRIPT_VERSION=/ c\SCRIPT_VERSION=\"$VERSION\"" ./weave
-    sed -i -e "s/:latest/:$VERSION/" -e "/imagePullPolicy: Always/d" ./prog/weave-kube/weave-daemonset.yaml
-    sed -i -e "s/:latest/:$VERSION/" -e "/imagePullPolicy: Always/d" ./prog/weave-kube/weave-daemonset-k8s-1.6.yaml
+    sed -i.bak "/SCRIPT_VERSION=/ c\SCRIPT_VERSION=\"$VERSION\"" ./weave
+    sed -i.bak -e "s/:latest/:$VERSION/" -e "/imagePullPolicy: Always/d" ./prog/weave-kube/weave-daemonset.yaml
+    sed -i.bak -e "s/:latest/:$VERSION/" -e "/imagePullPolicy: Always/d" ./prog/weave-kube/weave-daemonset-k8s-1.6.yaml
+    sed -i.bak -e "s/:latest/:$VERSION/" -e "/imagePullPolicy: Always/d" ./prog/weave-kube/weave-daemonset-k8s-1.7.yaml
     make SUDO=$SUDO WEAVE_VERSION=$VERSION DOCKERHUB_USER=$DOCKERHUB_USER
 
     if make tests; then
@@ -173,6 +174,13 @@ draft() {
         --name "weave-daemonset-k8s-1.6.yaml" \
         --file "./prog/weave-kube/weave-daemonset-k8s-1.6.yaml"
 
+    github-release upload \
+        --user $GITHUB_USER \
+        --repo weave \
+        --tag $LATEST_TAG \
+        --name "weave-daemonset-k8s-1.7.yaml" \
+        --file "./prog/weave-kube/weave-daemonset-k8s-1.7.yaml"
+
     echo "** Draft $TYPE $RELEASE_NAME $VERSION created at"
     echo -e "\thttps://github.com/$GITHUB_USER/weave/releases/$LATEST_TAG"
 }
@@ -187,6 +195,10 @@ publish() {
     fi
 
     if [ "$RELEASE_TYPE" = 'PRERELEASE' ] ; then
+        echo "== Building and pushing plugin"
+        make UPDATE_LATEST=false SUDO=$SUDO WEAVE_VERSION=$VERSION DOCKERHUB_USER=$DOCKERHUB_USER plugin_publish
+        echo "** Plugin built and pushed"
+
         echo "== Tagging and pushing images on docker hub as user $DOCKERHUB_USER"
         make UPDATE_LATEST=$IS_MAINLINE SUDO=$SUDO WEAVE_VERSION=$VERSION DOCKERHUB_USER=$DOCKERHUB_USER publish
         echo "** Docker images tagged and pushed"
@@ -219,6 +231,10 @@ publish() {
         fi
         echo '** Sanity checks OK for publishing tag' $LATEST_TAG as $DOCKERHUB_USER/weave:$VERSION
 
+        echo "== Building and pushing plugin"
+        make UPDATE_LATEST=true SUDO=$SUDO WEAVE_VERSION=$VERSION DOCKERHUB_USER=$DOCKERHUB_USER plugin_publish
+        echo "** Plugin built and pushed"
+
         echo "== Tagging and pushing images on docker hub as user $DOCKERHUB_USER"
         make PUBLISH_WEAVEDB=$IS_MAINLINE UPDATE_LATEST=$IS_MAINLINE SUDO=$SUDO WEAVE_VERSION=$VERSION DOCKERHUB_USER=$DOCKERHUB_USER publish
         echo "** Docker images tagged and pushed"
@@ -266,6 +282,13 @@ publish() {
             --name "weave-daemonset-k8s-1.6.yaml" \
             --file "./prog/weave-kube/weave-daemonset-k8s-1.6.yaml"
 
+        github-release upload \
+            --user $GITHUB_USER \
+            --repo weave \
+            --tag latest_release \
+            --name "weave-daemonset-k8s-1.7.yaml" \
+            --file "./prog/weave-kube/weave-daemonset-k8s-1.7.yaml"
+
         echo "** Release $RELEASE_NAME $VERSION published at"
         echo -e "\thttps://github.com/$GITHUB_USER/weave/releases/$LATEST_TAG"
         echo -e "\thttps://github.com/$GITHUB_USER/weave/releases/latest_release"

build/Dockerfile

@@ -1,24 +1,27 @@
-FROM golang:1.7.4
+FROM golang:1.8.1
 
 # Support Raspberry Pi 2 and newer
 ENV GOARM 7
 
-# The names of the architectures that we're fetching cross-compilers for from the Ubuntu packages
+# The names of the architectures that we're fetching cross-compilers for from the Debian packages
 ENV DEB_CROSSPLATFORMS \
 	armhf \
-	arm64
+	arm64 \
+	ppc64el
 
 # The go platforms we are supporting to cross-compile
 ENV GO_CROSSPLATFORMS \
 	linux/arm \
-	linux/arm64
+	linux/arm64 \
+	linux/ppc64le
 
 # The names of the gcc cross-compilers we have installed
 ENV GCC_CROSSCOMPILERS \
 	arm-linux-gnueabihf \
-	aarch64-linux-gnu
+	aarch64-linux-gnu \
+	powerpc64le-linux-gnu
 
-# Temporarily add the Ubuntu repositories, because we're going to install gcc cross-compilers from there
+# Temporarily add the Debian repositories, because we're going to install gcc cross-compilers from there
 # Install the build-essential and crossbuild-essential-ARCH packages
 RUN echo "deb http://emdebian.org/tools/debian/ jessie main" > /etc/apt/sources.list.d/cgocrosscompiling.list \
   && curl http://emdebian.org/tools/debian/emdebian-toolchain-archive.key | apt-key add - \

build/shasums/docker-tgz-armel.sha256sum

@@ -1 +0,0 @@
-8f313a9df867c207ae9ceac6e0935d1d2a0360aafce780dd5d1e0687ee98e0c3 docker-1.10.3.tgz

build/shasums/docker-tgz-x86_64.sha256sum

@@ -1 +0,0 @@
-7541b5490cee1183d911f909c0e91d3d9d57b2b775e40f14a1a7a6de216ca4a8 docker-1.10.3.tgz

build/shasums/qemu-ppc64le-static.sha256sum

@@ -0,0 +1 @@
+926f7f265332076ea4abf6e9586288376dbed2322c155b8d0cafbee614aaa7f2  qemu-ppc64le-static

code-of-conduct.md

@@ -0,0 +1,7 @@
+## Community Code of Conduct
+
+Weaveworks follows the [CNCF Code of Conduct](https://github.com/cncf/foundation/blob/master/code-of-conduct.md).
+
+Instances of abusive, harassing, or otherwise unacceptable behavior
+may be reported by contacting a Weaveworks project maintainer, or
+Alexis Richardson alexis@weave.works.

common/docker/client.go

@@ -165,13 +165,25 @@ func (pending pendingStarts) finish(id string) {
 	}
 }
 
+// AllContainerIDs returns all the IDs of Docker containers,
+// whether they are running or not.
 func (c *Client) AllContainerIDs() ([]string, error) {
-	all, err := c.ListContainers(docker.ListContainersOptions{All: true})
+	return c.containerIDs(true)
+}
+
+// RunningContainerIDs returns all the IDs of the running
+// Docker containers.
+func (c *Client) RunningContainerIDs() ([]string, error) {
+	return c.containerIDs(false)
+}
+
+func (c *Client) containerIDs(all bool) ([]string, error) {
+	containers, err := c.ListContainers(docker.ListContainersOptions{All: all})
 	if err != nil {
 		return nil, err
 	}
 	var ids []string
-	for _, c := range all {
+	for _, c := range containers {
 		ids = append(ids, c.ID)
 	}
 	return ids, nil
@@ -213,6 +225,30 @@ func (c *Client) GetContainerIP(nameOrID string) (string, error) {
 	return info.NetworkSettings.IPAddress, nil
 }
 
+func (c *Client) EnsureNetwork(networkName, driver, subnet string, options map[string]interface{}) error {
+	_, err := c.CreateNetwork(
+		docker.CreateNetworkOptions{
+			Name:           networkName,
+			CheckDuplicate: true,
+			Driver:         driver,
+			IPAM: docker.IPAMOptions{
+				Driver: driver,
+				Config: []docker.IPAMConfig{{Subnet: subnet}},
+			},
+			Options: options,
+		})
+	if err != docker.ErrNetworkAlreadyExists && err != nil {
+		// Despite appearances to the contrary, CreateNetwork does
+		// sometimes(always?) *not* return ErrNetworkAlreadyExists
+		// when the network already exists. Hence we need to check for
+		// this explicitly.
+		if _, err2 := c.NetworkInfo(networkName); err2 != nil {
+			return fmt.Errorf("unable to create network: %s", err)
+		}
+	}
+	return nil
+}
+
 // logging
 
 func (c *Client) errorf(fmt string, args ...interface{}) {

common/wait.go

@@ -0,0 +1,25 @@
+package common
+
+import (
+	"sync"
+	"sync/atomic"
+)
+
+type WaitGroup struct {
+	wg    sync.WaitGroup
+	count int32
+}
+
+// Increase the count of things waiting and return a closure which calls Done
+func (w *WaitGroup) Add() func() {
+	atomic.AddInt32(&w.count, 1)
+	w.wg.Add(1)
+	return func() {
+		w.wg.Done()
+		atomic.AddInt32(&w.count, -1)
+	}
+}
+
+func (w *WaitGroup) IsDone() bool {
+	return atomic.LoadInt32(&w.count) == 0
+}

db/boltdb.go

@@ -26,10 +26,15 @@ var (
 
 const (
 	NameIdent = "peername"
-	FileName  = "data.db"
+	fileName  = "data.db"
 )
 
-func NewBoltDB(dbPathname string) (*BoltDB, error) {
+func Pathname(dbPrefix string) string {
+	return dbPrefix + fileName
+}
+
+func NewBoltDB(dbPrefix string) (*BoltDB, error) {
+	dbPathname := Pathname(dbPrefix)
 	db, err := bolt.Open(dbPathname, 0660, nil)
 	if err != nil {
 		return nil, fmt.Errorf("[boltDB] Unable to open %s: %s", dbPathname, err)
@@ -38,8 +43,9 @@ func NewBoltDB(dbPathname string) (*BoltDB, error) {
 	return &BoltDB{db: db}, err
 }
 
-func NewBoltDBReadOnly(dbPathname string) (*BoltDB, error) {
+func NewBoltDBReadOnly(dbPrefix string) (*BoltDB, error) {
 	options := bolt.Options{Timeout: time.Millisecond * 50, ReadOnly: true}
+	dbPathname := Pathname(dbPrefix)
 	db, err := bolt.Open(dbPathname, 0660, &options)
 	if err != nil {
 		return nil, fmt.Errorf("[boltDB] Unable to open %s: %s", dbPathname, err)

docs/benchmarks.md

@@ -0,0 +1,81 @@
+# Weave Net Benchmarks
+
+## 2017-02-16: Fast Datapath Encryption
+
+Results: https://www.weave.works/weave-net-performance-fast-datapath/
+
+* Two `c3.8xlarge` instances running Ubuntu 16.04 LTS (`ami-d8f4deab`) in
+`eu-west-1` region of AWS.
+* 10 Gigabit [Enhanced Networking](http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/enhanced-networking.html).
+* `iperf 3.0.11`.
+
+### fastdp
+
+* Encrypted:
+
+```
+host1$ WEAVE_MTU=8916 weave launch --password=foobar
+host1$ docker $(weave config) run --rm -ti --name server networkstatic/iperf3 -s
+
+host2$ WEAVE_MTU=8916 weave launch --password=foobar host1
+host2$ docker $(weave config) run --rm -ti networkstatic/iperf3 -c server
+```
+
+* Non-encrypted:
+
+```
+host1$ WEAVE_MTU=8950 weave launch
+host1$ docker $(weave config) run --rm -ti --name server networkstatic/iperf3 -s
+
+host2$ WEAVE_MTU=8950 weave launch --password=foobar host1
+host2$ docker $(weave config) run --rm -ti networkstatic/iperf3 -c server
+```
+
+### sleeve
+
+* Encrypted:
+
+```
+host1$ WEAVE_NO_FASTDP=1 weave launch --password=foobar
+host1$ docker $(weave config) run --rm -ti --name server networkstatic/iperf3 -s
+
+host2$ WEAVE_NO_FASTDP=1 weave launch --password=foobar host1
+host2$ docker $(weave config) run --rm -ti networkstatic/iperf3 -c server
+```
+
+* Non-encrypted:
+
+```
+host1$ WEAVE_NO_FASTDP=1 weave launch
+host1$ docker $(weave config) run --rm -ti --name server networkstatic/iperf3 -s
+
+host2$ WEAVE_NO_FASTDP=1 weave launch --password=foobar host1
+host2$ docker $(weave config) run --rm -ti networkstatic/iperf3 -c server
+```
+
+### host
+
+* Non-encrypted:
+
+```
+host1$ iperf -s
+host2$ iperf -c host1
+```
+
+* Encrypted:
+
+```
+host1$ export KEY1="0x466454f2b1a770f8f872f9afbc35ebeac57e00fc11ac86ed1f82716f010b20f0cf532274"
+host1$ export KEY2="0x3531151241a770f8f872f9afbc35ebeac57e00fc11ac86ed1f82716f010b20f0cf532274"
+host1# ip xfrm state add src ${IP1} dst ${IP2} proto esp spi 0x4c856ffc replay-window 256 flag esn reqid 0 mode transport aead 'rfc4106(gcm(aes))' ${KEY1} 128
+host1# ip xfrm state add src ${IP2} dst ${IP1} proto esp spi 0x9b0830bc replay-window 256 flag esn reqid 0 mode transport aead 'rfc4106(gcm(aes))' ${KEY2} 128
+host1# ip xfrm policy add src ${IP1}/32 dst ${IP2}/32 dir out tmpl src ${IP1} dst ${IP2} proto esp spi 0x4c856ffc reqid 0 mode transport
+host1$ iperf -s
+
+host2$ export KEY1="0x466454f2b1a770f8f872f9afbc35ebeac57e00fc11ac86ed1f82716f010b20f0cf532274"
+host2$ export KEY2="0x3531151241a770f8f872f9afbc35ebeac57e00fc11ac86ed1f82716f010b20f0cf532274"
+host2# ip xfrm state add src ${IP2} dst ${IP1} proto esp spi 0x9b0830bc reqid 0 replay-window 256 flag esn mode transport aead 'rfc4106(gcm(aes))' ${KEY2} 128
+host2# ip xfrm state add src ${IP1} dst ${IP2} proto esp spi 0x4c856ffc reqid 0 replay-window 256 flag esn mode transport aead 'rfc4106(gcm(aes))' ${KEY1} 128
+host2# ip xfrm policy add src ${IP2}/32 dst ${IP1}/32 dir out tmpl src ${IP1} dst ${IP2} proto esp spi 0x9b0830bc reqid 0 mode transport
+host2$ iperf -c host1
+```

docs/entrypoints.md

@@ -0,0 +1,70 @@
+# Entrypoints
+
+This document describes the various ways that weave as a binary gets launched, and where and how it processes options.
+
+The goal is to enable maintainers to modify functionality without having to reverse engineer the code to find all of the places you can start a weave binary and what it depends upon.
+
+## Components
+### weaver
+
+`weaver` is the primary binary: it sets up the weave bridge to which containers are attached, manages IP address allocation, serves DNS requests and does many more things.
+
+Since version 2.0, `weaver` also bundles the Docker Plugin (legacy version) and Docker API Proxy which used to be shipped as separate containers.
+
+### weaveutil
+weaveutil is a binary which provides a number of functions for managing an existing weave network. It gets information about the network, attaches and detaches containers, etc.
+
+For the majority of interactions with an existing weave network, you will launch `weaveutil` in some manner or another.
+
+Almost all options can be passed as `--option` to `weaveutil`. These options are created in `prog/weaveutil/main.go` with a table mapping each command to a dedicated golang function.
+
+In addition to operating in normal mode, `weaveutil` has several additional operating modes. Before processing any commands, `weaveutil` checks the filename with which it was called.
+
+* If it was `weave-ipam` it delegates responsibility to the `cniIPAM()` function in `prog/weaveutil/cni.go`, which, in turn, calls the standard CNI plugin function `cni.PluginMain()`, passing it weave's IPAM implementation from `plugin/ipam`.
+* If it was `weave-net` it delegates responsibility to the `cniNet()` function in `prog/weaveutil/cni.go`, which, in turn, calls the standard CNI plugin function `cni.PluginMain()`, passing it weave's net plugin implementation from `plugin/net`.
+
+Finally, `weaveutil` is called by `weaver` when it needs to operate in a different network namespace.
+Go programs cannot safely switch namespaces; see this [issue](http://github.com/vishvananda/netns/issues/17).
+The call chain is `weaver`->`nsenter`->`weaveutil`.
+
+### weave
+Wrapping `weaver` and `weaveutil` is `weave`, a `sh` script that provides help information and calls `weaveutil` as relevant.
+
+### weave-kube
+weave-kube is an image with the weave binaries and a wrapper script installed. It is responsible for:
+
+1. Setting up the weave network
+2. Connecting to peers
+3. Copy the weave CNI-compatible binary plugin `weaveutil` to `/opt/cni/bin/weave-net` and weave config in `/etc/cni/net.d/10-weave.conf`
+4. Running the weave network policy controller (weave-npc) to implement kubernetes' `NetworkPolicy`
+
+Once installation is complete, each network-relevant change in a container leads `kubelet` to:
+
+1. Set certain environment variables to configure the CNI plugin, mostly related to the container ID
+2. Launch `/opt/cni/bin/weave-net`
+3. Pass the CNI config - the contents of `/etc/cni/net.d/10-weave.conf` to `weave-net`
+
+Thus, when each container is changed, and `kubelet` calls weave as a CNI plugin, it really just is launching `weaveutil` as `weave-net`.
+
+The installation and setup of all of the above - and therefore the entrypoint to the weave-kube image - is the script `prog/weave-kube/launch.sh`. `launch.sh` does the following:
+
+1. Read configuration variables from the environment. When the documentation for `weave-kube` describes configuring the weave network by changing the environment variables in the daemonset in the `.yml` file, `launch.sh` reads these environment variables.
+2. Set up the config file at `/etc/cni/net.d/10-weave.conf`
+3. Run the `weave` initialization script
+4. Run `weaver` with the correct configuration passed as command-line options
+
+
+## Adding Options
+To add new options to how weave should run with each invocation, you would do the following:
+
+1. determine to which `weave` command you want to add the option(s). `weave` normally is launched as `weave <command> <options> ...`.
+2. Add the option to `weave` script help for each `weave` command you wish to make it available. As of this writing, all of the commands and their options are listed in the function `usage_no_exist()`.
+3. Add the option to `weave` script option processing for the `weave` command, under the `case $COMMAND in` in the main `weave` script.
+4. In the function for the `weave` command, determine if the command should be passed on to `weaveutil` or `weaver`. Pass the option on to `weaveutil` or `weaver`, as appropriate, in the format `--option`, in the function for the `weave` command.
+5. Add a command-line option `--option` to `weaveutil` or `weaver` as appropriate.
+6. If the option can be configured for CNI:
+    * have `prog/weave-kube/launch.sh` read it as an environment variable and set inside
+    * set a default for the environment variable in `prog/weave-kube/weave-daemonset-k8s-1.6.yaml` and `weave-daemonset.yaml`
+    * if it should be set via `weaver` globally on its one-time initialization invocation, pass it on in `launch.sh`
+    * if it needs to be set via `weaveutil` on each invocation of `weave-net`, have `launch.sh` save it as an option in the CNI config file `/etc/cni/net.d/10-weave.conf` and then have the CNI code in `plugin/net/cni.go` in `weaveutil` read it and use where appropriate
+7. Document it!

docs/release-process.md

@@ -4,8 +4,7 @@
 * Up-to-date versions of `git`, `go` etc
 * Install GitHub release tool `go get github.com/weaveworks/github-release`
 * Create a [github token for
-  github-release](https://help.github.com/articles/creating-an-access-token-for-command-line-use/);
-set and export `$GITHUB_TOKEN` with this value
+  github-release](https://help.github.com/articles/creating-an-access-token-for-command-line-use/); select the `repo` OAuth scope; set and export `$GITHUB_TOKEN` with this value
 
 ## Release Types
 

ipam/allocator_test.go

@@ -9,6 +9,7 @@ import (
 
 	"github.com/stretchr/testify/require"
 
+	"github.com/weaveworks/weave/api"
 	"github.com/weaveworks/weave/common"
 	"github.com/weaveworks/weave/net/address"
 	"github.com/weaveworks/weave/testing/gossip"
@@ -30,6 +31,10 @@ func (alloc *Allocator) SimplyClaim(ident string, cidr address.CIDR) error {
 	return alloc.Claim(ident, cidr, true, true, returnFalse)
 }
 
+func (alloc *Allocator) AnonymousClaim(cidr address.CIDR) error {
+	return alloc.Claim(api.NoContainerID, cidr, false, true, returnFalse)
+}
+
 func TestAllocFree(t *testing.T) {
 	const (
 		container1 = "abcdef"
@@ -170,6 +175,8 @@ func TestAllocatorClaim(t *testing.T) {
 		universe   = "10.0.3.0/24"
 		testAddr1  = "10.0.3.5/24"
 		testAddr2  = "10.0.4.2/24"
+		testAddr4  = "10.0.2.3/23"
+		subnet4    = "10.0.2.0/23"
 		testPre    = "10.0.3.1/24"
 	)
 
@@ -203,6 +210,9 @@ func TestAllocatorClaim(t *testing.T) {
 	// claim for a different container should fail
 	err = alloc.SimplyClaim(container1, addr1)
 	require.Error(t, err)
+	// claim for an anonymous container, will assume it's the same one
+	err = alloc.AnonymousClaim(addr1)
+	require.NoError(t, err)
 	// claiming the address allocated on the other peer should fail
 	err = alloc.SimplyClaim(container1, address.MakeCIDR(subnet, addrx))
 	require.Error(t, err, "claiming address allocated on other peer should fail")
@@ -215,6 +225,15 @@ func TestAllocatorClaim(t *testing.T) {
 	addr2, _ := address.ParseCIDR(testAddr2)
 	err = alloc.SimplyClaim(container1, addr2)
 	require.NoError(t, err)
+	err = alloc.SimplyClaim(container3, addr2)
+	require.Error(t, err)
+	// Anonymous claim outside of our universe
+	addr4, _ := address.ParseCIDR(testAddr4)
+	subnet4cidr, _ := address.ParseCIDR(subnet4)
+	require.NoError(t, alloc.AnonymousClaim(addr4))
+	addr5, err := alloc.Allocate(api.NoContainerID, subnet4cidr, false, returnFalse)
+	require.NoError(t, err)
+	require.NotEqual(t, addr4.Start(), addr5)
 }
 
 func (alloc *Allocator) pause() func() {

ipam/claim.go

@@ -40,10 +40,29 @@ func (c *claim) Try(alloc *Allocator) bool {
 		return true
 	}
 
+	addOwned := func() {
+		if c.ident == api.NoContainerID {
+			alloc.addOwned(c.cidr.Addr.String(), c.cidr, c.isContainer)
+		} else {
+			alloc.addOwned(c.ident, c.cidr, c.isContainer)
+		}
+	}
+
 	if !alloc.ring.Contains(c.cidr.Addr) {
-		// Address not within our universe; assume user knows what they are doing
 		alloc.infof("Address %s claimed by %s - not in our range", c.cidr, c.ident)
-		alloc.addOwned(c.ident, c.cidr, c.isContainer)
+		previousOwner := alloc.findOwner(c.cidr.Addr)
+		switch {
+		case previousOwner == "":
+			addOwned()
+		case previousOwner == c.ident: // already owned by this ID
+		case c.ident == api.NoContainerID: // already owned by anonymous container
+			// do nothing (no automatic fall-through in Go)
+		case !alloc.dead[previousOwner].IsZero(): // already owned by dead container
+			alloc.removeOwned(previousOwner, c.cidr.Addr)
+			addOwned()
+		default:
+			c.sendResult(fmt.Errorf("address %s already in use by %s", c.cidr, previousOwner))
+		}
 		c.sendResult(nil)
 		return true
 	}
@@ -85,11 +104,7 @@ func (c *claim) Try(alloc *Allocator) bool {
 		// Unused address, we try to claim it:
 		if err := alloc.space.Claim(c.cidr.Addr); err == nil {
 			alloc.debugln("Claimed", c.cidr, "for", c.ident)
-			if c.ident == api.NoContainerID {
-				alloc.addOwned(c.cidr.Addr.String(), c.cidr, c.isContainer)
-			} else {
-				alloc.addOwned(c.ident, c.cidr, c.isContainer)
-			}
+			addOwned()
 			c.sendResult(nil)
 		} else {
 			c.sendResult(err)

ipam/http.go

@@ -1,6 +1,7 @@
 package ipam
 
 import (
+	"encoding/json"
 	"fmt"
 	"net/http"
 
@@ -127,6 +128,39 @@ func (alloc *Allocator) HandleHTTP(router *mux.Router, defaultSubnet address.CID
 		writeAddresses(w, addrs)
 	})
 
+	router.Methods("GET").Path("/ip").HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		type mapping struct {
+			ContainerID string   `json:"containerid"`
+			Addrs       []string `json:"addrs"`
+		}
+
+		type mappings struct {
+			Owned []mapping `json:"owned"`
+		}
+
+		resultChan := make(chan mappings)
+		alloc.actionChan <- func() {
+			ms := mappings{}
+			for containerid, d := range alloc.owned {
+				m := mapping{
+					ContainerID: containerid,
+					Addrs:       []string{},
+				}
+				for _, addr := range d.Cidrs {
+					m.Addrs = append(m.Addrs, addr.String())
+				}
+				ms.Owned = append(ms.Owned, m)
+			}
+			resultChan <- ms
+		}
+		ms := <-resultChan
+
+		w.Header().Set("Content-Type", "application/json")
+		if err := json.NewEncoder(w).Encode(ms); err != nil {
+			common.Log.Warningln("[allocator]:", err.Error())
+		}
+	})
+
 	router.Methods("POST").Path("/ip/{id}/{ip}/{prefixlen}").HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		vars := mux.Vars(r)
 		if subnet, ok := parseCIDR(w, vars["ip"]+"/"+vars["prefixlen"], true); ok {
@@ -164,17 +198,6 @@ func (alloc *Allocator) HandleHTTP(router *mux.Router, defaultSubnet address.CID
 		w.WriteHeader(204)
 	})
 
-	router.Methods("DELETE").Path("/peer").HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
-		alloc.Shutdown()
-		w.WriteHeader(204)
-	})
-
-	router.Methods("DELETE").Path("/peer/{id}").HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
-		ident := mux.Vars(r)["id"]
-		transferred := alloc.AdminTakeoverRanges(ident)
-		fmt.Fprintf(w, "%d IPs taken over from %s\n", transferred, ident)
-	})
-
 	router.Methods("GET").Path("/ipinfo/tracker").HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		fmt.Fprintf(w, tracker)
 	})

ipam/tracker/awsvpc.go

@@ -30,7 +30,6 @@ import (
 	"github.com/vishvananda/netlink"
 
 	"github.com/weaveworks/weave/common"
-	wnet "github.com/weaveworks/weave/net"
 	"github.com/weaveworks/weave/net/address"
 )
 
@@ -42,7 +41,7 @@ type AWSVPCTracker struct {
 }
 
 // NewAWSVPCTracker creates and initialises AWS VPC based tracker.
-func NewAWSVPCTracker() (*AWSVPCTracker, error) {
+func NewAWSVPCTracker(bridgeName string) (*AWSVPCTracker, error) {
 	var (
 		err     error
 		session = session.New()
@@ -69,9 +68,9 @@ func NewAWSVPCTracker() (*AWSVPCTracker, error) {
 	t.routeTableID = *routeTableID
 
 	// Detect Weave bridge link index
-	link, err := netlink.LinkByName(wnet.WeaveBridgeName)
+	link, err := netlink.LinkByName(bridgeName)
 	if err != nil {
-		return nil, fmt.Errorf("cannot find \"%s\" interface: %s", wnet.WeaveBridgeName, err)
+		return nil, fmt.Errorf("cannot find \"%s\" interface: %s", bridgeName, err)
 	}
 	t.linkIndex = link.Attrs().Index
 

nameserver/http.go

@@ -27,7 +27,7 @@ func (n *Nameserver) HandleHTTP(router *mux.Router, dockerCli *docker.Client) {
 			vars      = mux.Vars(r)
 			container = vars["container"]
 			ipStr     = vars["ip"]
-			hostname  = dns.Fqdn(r.FormValue("fqdn"))
+			fqdn      = r.FormValue("fqdn")
 			ip, err   = address.ParseIP(ipStr)
 		)
 		if err != nil {
@@ -35,16 +35,11 @@ func (n *Nameserver) HandleHTTP(router *mux.Router, dockerCli *docker.Client) {
 			return
 		}
 
-		if !dns.IsSubDomain(n.domain, hostname) {
-			n.infof("Ignoring registration %s %s %s (not a subdomain of %s)", hostname, ipStr, container, n.domain)
-			return
-		}
-
-		n.AddEntry(hostname, container, n.ourName, ip)
+		n.AddEntryFQDN(fqdn, container, n.ourName, ip)
 
 		if r.FormValue("check-alive") == "true" && dockerCli != nil && dockerCli.IsContainerNotRunning(container) {
 			n.infof("container '%s' is not running: removing", container)
-			n.Delete(hostname, container, ipStr, ip)
+			n.Delete(dns.Fqdn(fqdn), container, ipStr, ip)
 		}
 
 		w.WriteHeader(204)

nameserver/nameserver.go

@@ -91,6 +91,15 @@ func (n *Nameserver) AddEntry(hostname, containerid string, origin mesh.PeerName
 	n.broadcastEntries(entry)
 }
 
+func (n *Nameserver) AddEntryFQDN(fqdn, containerid string, origin mesh.PeerName, addr address.Address) {
+	hostname := dns.Fqdn(fqdn)
+	if !dns.IsSubDomain(n.domain, hostname) {
+		n.infof("Ignoring registration %s %s %s (not a subdomain of %s)", hostname, addr.String(), containerid, n.domain)
+		return
+	}
+	n.AddEntry(hostname, containerid, origin, addr)
+}
+
 func (n *Nameserver) Lookup(hostname string) []address.Address {
 	n.RLock()
 	defer n.RUnlock()

net/address/address.go

@@ -160,6 +160,11 @@ func (cidr CIDR) String() string {
 	return fmt.Sprintf("%s/%d", cidr.Addr.String(), cidr.PrefixLen)
 }
 
+func (cidr CIDR) IPNet() (r *net.IPNet) {
+	mask := net.CIDRMask(cidr.PrefixLen, 8*net.IPv4len)
+	return &net.IPNet{IP: cidr.Addr.IP4(), Mask: mask}
+}
+
 // FromIP4 converts an ipv4 address to our integer address type
 func FromIP4(ip4 net.IP) (r Address) {
 	for _, b := range ip4.To4() {

net/address/address_test.go

@@ -1,6 +1,7 @@
 package address
 
 import (
+	"net"
 	"testing"
 	"testing/quick"
 
@@ -67,3 +68,15 @@ func TestCIDRStartAndEnd(t *testing.T) {
 	require.Equal(t, ip("10.0.0.0"), cidr.Start(), "")
 	require.Equal(t, ip("10.0.1.0"), cidr.End(), "")
 }
+
+func TestConversions(t *testing.T) {
+	addrStr := "10.2.3.0"
+	cidrStr := addrStr + "/24"
+	addr := ip(addrStr)
+	require.Equal(t, addr.String(), addrStr)
+	require.Equal(t, addr.IP4(), net.ParseIP(addrStr).To4())
+	cidr, _ := ParseCIDR(cidrStr)
+	require.Equal(t, cidr.String(), cidrStr)
+	_, ipnet, _ := net.ParseCIDR(cidrStr)
+	require.Equal(t, cidr.IPNet(), ipnet)
+}

net/arp.go

@@ -1,27 +1,27 @@
 package net
 
-import "fmt"
-import "io"
-import "os"
+import (
+	"fmt"
+	"io"
+	"os"
+	"path/filepath"
+)
 
 // Configure the ARP cache parameters for the given interface.  This
 // makes containers react more quickly to a change in the MAC address
 // associated with an IP address.
-func ConfigureARPCache(name string) error {
-	if err := sysctl(fmt.Sprintf("net/ipv4/neigh/%s/base_reachable_time", name), "5"); err != nil {
+func ConfigureARPCache(procPath, name string) error {
+	if err := sysctl(procPath, fmt.Sprintf("net/ipv4/neigh/%s/base_reachable_time", name), "5"); err != nil {
 		return err
 	}
-	if err := sysctl(fmt.Sprintf("net/ipv4/neigh/%s/delay_first_probe_time", name), "2"); err != nil {
+	if err := sysctl(procPath, fmt.Sprintf("net/ipv4/neigh/%s/delay_first_probe_time", name), "2"); err != nil {
 		return err
 	}
-	if err := sysctl(fmt.Sprintf("net/ipv4/neigh/%s/ucast_solicit", name), "1"); err != nil {
-		return err
-	}
-	return nil
+	return sysctl(procPath, fmt.Sprintf("net/ipv4/neigh/%s/ucast_solicit", name), "1")
 }
 
-func sysctl(variable, value string) error {
-	f, err := os.OpenFile(fmt.Sprintf("/proc/sys/%s", variable), os.O_WRONLY, 0)
+func sysctl(procPath, variable, value string) error {
+	f, err := os.OpenFile(filepath.Join(procPath, "/sys/", variable), os.O_WRONLY, 0)
 	if err != nil {
 		return err
 	}

net/bridge.go

@@ -2,43 +2,136 @@ package net
 
 import (
 	"fmt"
+	"io/ioutil"
 	"net"
+	"path/filepath"
 
 	"github.com/Sirupsen/logrus"
+	"github.com/coreos/go-iptables/iptables"
+	"github.com/pkg/errors"
 	"github.com/vishvananda/netlink"
+
+	"github.com/weaveworks/weave/common/odp"
 )
 
-type BridgeType int
+/* This code implements three possible configurations to connect
+   containers to the Weave Net overlay:
+
+1. Bridge
+                 +-------+
+(container-veth)-+ weave +-(vethwe-bridge)--(vethwe-pcap)
+                 +-------+
+
+"weave" is a Linux bridge. "vethwe-pcap" (end of veth pair) is used
+to capture and inject packets, by router/pcap.go.
+
+2. BridgedFastdp
+
+                 +-------+                                    /----------\
+(container-veth)-+ weave +-(vethwe-bridge)--(vethwe-datapath)-+ datapath +
+                 +-------+                                    \----------/
+
+"weave" is a Linux bridge and "datapath" is an Open vSwitch datapath;
+they are connected via a veth pair. Packet capture and injection use
+the "datapath" device, via "router/fastdp.go:fastDatapathBridge"
+
+3. Fastdp
+
+                 /-------\
+(container-veth)-+ weave +
+                 \-------/
+
+"weave" is an Open vSwitch datapath, and capture/injection are as in
+BridgedFastdp. Not used by default due to missing conntrack support in
+datapath of old kernel versions (https://github.com/weaveworks/weave/issues/1577).
+*/
 
 const (
 	WeaveBridgeName = "weave"
 	DatapathName    = "datapath"
-
-	None BridgeType = iota
-	Bridge
-	Fastdp
-	BridgedFastdp
-	Inconsistent
+	DatapathIfName  = "vethwe-datapath"
+	BridgeIfName    = "vethwe-bridge"
+	PcapIfName      = "vethwe-pcap"
 )
 
-func DetectBridgeType(weaveBridgeName, datapathName string) BridgeType {
+type Bridge interface {
+	init(config *BridgeConfig) error // create and initialise bridge device(s)
+	attach(veth *netlink.Veth) error // attach veth to bridge
+	IsFastdp() bool                  // does this bridge use fastdp?
+	String() string                  // human-readable type string
+}
+
+// Used to indicate a fallback to the Bridge type
+var errBridgeNotSupported = errors.New("bridge not supported")
+
+type bridgeImpl struct{ bridge netlink.Link }
+type fastdpImpl struct{ datapathName string }
+type bridgedFastdpImpl struct {
+	bridgeImpl
+	fastdpImpl
+}
+
+// Returns a string that is consistent with the weave script
+func (bridgeImpl) String() string        { return "bridge" }
+func (fastdpImpl) String() string        { return "fastdp" }
+func (bridgedFastdpImpl) String() string { return "bridged_fastdp" }
+
+// Used to decide whether to manage ODP tunnels
+func (bridgeImpl) IsFastdp() bool        { return false }
+func (fastdpImpl) IsFastdp() bool        { return true }
+func (bridgedFastdpImpl) IsFastdp() bool { return true }
+
+func ExistingBridgeType(weaveBridgeName, datapathName string) (Bridge, error) {
 	bridge, _ := netlink.LinkByName(weaveBridgeName)
 	datapath, _ := netlink.LinkByName(datapathName)
 
 	switch {
 	case bridge == nil && datapath == nil:
-		return None
+		return nil, nil
 	case isBridge(bridge) && datapath == nil:
-		return Bridge
+		return bridgeImpl{bridge: bridge}, nil
 	case isDatapath(bridge) && datapath == nil:
-		return Fastdp
+		return fastdpImpl{datapathName: datapathName}, nil
 	case isDatapath(datapath) && isBridge(bridge):
-		return BridgedFastdp
+		return bridgedFastdpImpl{bridgeImpl{bridge: bridge}, fastdpImpl{datapathName: datapathName}}, nil
 	default:
-		return Inconsistent
+		return nil, errors.New("Inconsistent bridge state detected. Please do 'weave reset' and try again")
 	}
 }
 
+func EnforceAddrAssignType(bridgeName string) (setAddr bool, err error) {
+	sysctlFilename := filepath.Join("/sys/class/net/", bridgeName, "/addr_assign_type")
+	addrAssignType, err := ioutil.ReadFile(sysctlFilename)
+	if err != nil {
+		return false, errors.Wrapf(err, "reading %q", sysctlFilename)
+	}
+
+	// From include/uapi/linux/netdevice.h
+	// #define NET_ADDR_PERM       0   /* address is permanent (default) */
+	// #define NET_ADDR_RANDOM     1   /* address is generated randomly */
+	// #define NET_ADDR_STOLEN     2   /* address is stolen from other device */
+	// #define NET_ADDR_SET        3   /* address is set using dev_set_mac_address() */
+	// Note the file typically has a newline at the end, so we just look at the first char
+	if addrAssignType[0] != '3' {
+		link, err := netlink.LinkByName(bridgeName)
+		if err != nil {
+			return false, errors.Wrapf(err, "EnforceAddrAssignType finding bridge %s", bridgeName)
+		}
+
+		mac, err := RandomMAC()
+		if err != nil {
+			return false, errors.Wrap(err, "creating random MAC")
+		}
+
+		if err := netlink.LinkSetHardwareAddr(link, mac); err != nil {
+			return false, errors.Wrapf(err, "setting bridge %s address to %v", bridgeName, mac)
+		}
+		return true, nil
+	}
+
+	return false, nil
+}
+
 func isBridge(link netlink.Link) bool {
 	_, isBridge := link.(*netlink.Bridge)
 	return isBridge
@@ -108,3 +201,289 @@ func FindBridgeIP(bridgeName string, subnet *net.IPNet) (net.IP, error) {
 	// No subnet, or none in the required subnet; just return the first one
 	return netdev.CIDRs[0].IP, nil
 }
+
+type BridgeConfig struct {
+	DockerBridgeName string
+	WeaveBridgeName  string
+	DatapathName     string
+	NoFastdp         bool
+	NoBridgedFastdp  bool
+	AWSVPC           bool
+	NPC              bool
+	MTU              int
+	Mac              string
+	Port             int
+}
+
+func (config *BridgeConfig) configuredBridgeType() Bridge {
+	switch {
+	case config.NoFastdp:
+		return bridgeImpl{}
+	case config.NoBridgedFastdp:
+		return fastdpImpl{datapathName: config.WeaveBridgeName}
+	default:
+		return bridgedFastdpImpl{fastdpImpl: fastdpImpl{datapathName: config.DatapathName}}
+	}
+}
+
+func EnsureBridge(procPath string, config *BridgeConfig, log *logrus.Logger) (Bridge, error) {
+	bridgeType, err := ExistingBridgeType(config.WeaveBridgeName, config.DatapathName)
+	if bridgeType != nil || err != nil {
+		return bridgeType, err
+	}
+
+	bridgeType = config.configuredBridgeType()
+	for {
+		if err := bridgeType.init(config); err != nil {
+			if errors.Cause(err) == errBridgeNotSupported {
+				log.Warnf("Skipping bridge creation of %q due to: %s", bridgeType, err)
+				bridgeType = bridgeImpl{}
+				continue
+			}
+			return nil, err
+		}
+		break
+	}
+
+	if err := configureIPTables(config); err != nil {
+		return bridgeType, errors.Wrap(err, "configuring iptables")
+	}
+
+	if config.AWSVPC {
+		// Set proxy_arp on the bridge, so that it could accept packets destined
+		// to containers within the same subnet but running on remote hosts.
+		// Without it, exact routes on each container are required.
+		if err := sysctl(procPath, "net/ipv4/conf/"+config.WeaveBridgeName+"/proxy_arp", "1"); err != nil {
+			return bridgeType, errors.Wrap(err, "setting proxy_arp")
+		}
+		// Avoid delaying the first ARP request. Also, setting it to 0 avoids
+		// placing the request into a bounded queue as it can be seen:
+		// https://git.kernel.org/cgit/linux/kernel/git/stable/linux-stable.git/tree/net/ipv4/arp.c?id=refs/tags/v4.6.1#n819
+		if err := sysctl(procPath, "net/ipv4/neigh/"+config.WeaveBridgeName+"/proxy_delay", "0"); err != nil {
+			return bridgeType, errors.Wrap(err, "setting proxy_arp")
+		}
+	}
+
+	if err := linkSetUpByName(config.WeaveBridgeName); err != nil {
+		return bridgeType, err
+	}
+
+	if err := ConfigureARPCache(procPath, config.WeaveBridgeName); err != nil {
+		return bridgeType, errors.Wrapf(err, "configuring ARP cache on bridge %q", config.WeaveBridgeName)
+	}
+
+	return bridgeType, nil
+}
+
+func (b bridgeImpl) initPrep(config *BridgeConfig) error {
+	mac, err := net.ParseMAC(config.Mac)
+	if err != nil {
+		return errors.Wrapf(err, "parsing bridge MAC %q", config.Mac)
+	}
+
+	linkAttrs := netlink.NewLinkAttrs()
+	linkAttrs.Name = config.WeaveBridgeName
+	linkAttrs.HardwareAddr = mac
+	if config.MTU == 0 {
+		config.MTU = 65535
+	}
+	b.bridge = &netlink.Bridge{LinkAttrs: linkAttrs}
+	if err = netlink.LinkAdd(b.bridge); err != nil {
+		return errors.Wrapf(err, "creating bridge %q", config.WeaveBridgeName)
+	}
+	if err := netlink.LinkSetHardwareAddr(b.bridge, mac); err != nil {
+		return errors.Wrapf(err, "setting bridge %q mac %v", config.WeaveBridgeName, mac)
+	}
+	// Attempting to set the bridge MTU to a high value directly
+	// fails. Bridges take the lowest MTU of their interfaces. So
+	// instead we create a temporary interface with the desired MTU,
+	// attach that to the bridge, and then remove it again.
+	dummy := &netlink.Dummy{LinkAttrs: netlink.NewLinkAttrs()}
+	dummy.LinkAttrs.Name = "vethwedu"
+	if err = netlink.LinkAdd(dummy); err != nil {
+		return errors.Wrap(err, "creating dummy interface")
+	}
+	if err := netlink.LinkSetMTU(dummy, config.MTU); err != nil {
+		return errors.Wrapf(err, "setting dummy interface mtu to %d", config.MTU)
+	}
+	if err := netlink.LinkSetMasterByIndex(dummy, b.bridge.Attrs().Index); err != nil {
+		return errors.Wrap(err, "setting dummy interface master")
+	}
+	if err := netlink.LinkDel(dummy); err != nil {
+		return errors.Wrap(err, "deleting dummy interface")
+	}
+
+	return nil
+}
+
+func (b bridgeImpl) init(config *BridgeConfig) error {
+	if err := b.initPrep(config); err != nil {
+		return err
+	}
+	if _, err := CreateAndAttachVeth(BridgeIfName, PcapIfName, config.WeaveBridgeName, config.MTU, true, func(veth netlink.Link) error {
+		return netlink.LinkSetUp(veth)
+	}); err != nil {
+		return errors.Wrap(err, "creating pcap veth pair")
+	}
+	if err := EthtoolTXOff(config.WeaveBridgeName); err != nil {
+		return errors.Wrap(err, "setting tx off")
+	}
+
+	return nil
+}
+
+func (f fastdpImpl) init(config *BridgeConfig) error {
+	odpSupported, err := odp.CreateDatapath(f.datapathName)
+	if !odpSupported {
+		msg := ""
+		if err != nil {
+			msg = err.Error()
+		}
+		return errors.Wrap(errBridgeNotSupported, msg)
+	}
+	if err != nil {
+		return errors.Wrapf(err, "creating datapath %q", f.datapathName)
+	}
+	datapath, err := netlink.LinkByName(f.datapathName)
+	if err != nil {
+		return errors.Wrapf(err, "finding datapath %q", f.datapathName)
+	}
+	if config.MTU == 0 {
+		/* GCE has the lowest underlay network MTU we're likely to encounter on
+		   a local network, at 1460 bytes.  To get the overlay MTU from that we
+		   subtract 20 bytes for the outer IPv4 header, 8 bytes for the outer
+		   UDP header, 8 bytes for the vxlan header, and 14 bytes for the inner
+		   ethernet header.  In addition, we subtract 34 bytes for the ESP overhead
+		   which is needed for the vxlan encryption. */
+		config.MTU = 1376
+	}
+	if err := netlink.LinkSetMTU(datapath, config.MTU); err != nil {
+		return errors.Wrapf(err, "setting datapath %q mtu %d", f.datapathName, config.MTU)
+	}
+	return nil
+}
+
+func (bf bridgedFastdpImpl) init(config *BridgeConfig) error {
+	if err := bf.fastdpImpl.init(config); err != nil {
+		return err
+	}
+	if err := bf.bridgeImpl.initPrep(config); err != nil {
+		return err
+	}
+	if _, err := CreateAndAttachVeth(BridgeIfName, DatapathIfName, config.WeaveBridgeName, config.MTU, true, func(veth netlink.Link) error {
+		if err := netlink.LinkSetUp(veth); err != nil {
+			return errors.Wrapf(err, "setting link up on %q", veth.Attrs().Name)
+		}
+		if err := odp.AddDatapathInterface(bf.datapathName, veth.Attrs().Name); err != nil {
+			return errors.Wrapf(err, "adding interface %q to datapath %q", veth.Attrs().Name, bf.datapathName)
+		}
+		return nil
+	}); err != nil {
+		return errors.Wrap(err, "creating bridged fastdp veth pair")
+	}
+
+	return linkSetUpByName(bf.datapathName)
+}
+
+func (b bridgeImpl) attach(veth *netlink.Veth) error {
+	return netlink.LinkSetMasterByIndex(veth, b.bridge.Attrs().Index)
+}
+
+func (bf bridgedFastdpImpl) attach(veth *netlink.Veth) error {
+	return bf.bridgeImpl.attach(veth)
+}
+
+func (f fastdpImpl) attach(veth *netlink.Veth) error {
+	return odp.AddDatapathInterface(f.datapathName, veth.Attrs().Name)
+}
+
+func configureIPTables(config *BridgeConfig) error {
+	ipt, err := iptables.New()
+	if err != nil {
+		return errors.Wrap(err, "creating iptables object")
+	}
+	if config.DockerBridgeName != "" {
+		if config.WeaveBridgeName != config.DockerBridgeName {
+			err := ipt.Insert("filter", "FORWARD", 1, "-i", config.DockerBridgeName, "-o", config.WeaveBridgeName, "-j", "DROP")
+			if err != nil {
+				return err
+			}
+		}
+
+		dockerBridgeIP, err := FindBridgeIP(config.DockerBridgeName, nil)
+		if err != nil {
+			return err
+		}
+
+		// forbid traffic to the Weave port from other containers
+		if err = ipt.AppendUnique("filter", "INPUT", "-i", config.DockerBridgeName, "-p", "tcp", "--dst", dockerBridgeIP.String(), "--dport", fmt.Sprint(config.Port), "-j", "DROP"); err != nil {
+			return err
+		}
+		if err = ipt.AppendUnique("filter", "INPUT", "-i", config.DockerBridgeName, "-p", "udp", "--dst", dockerBridgeIP.String(), "--dport", fmt.Sprint(config.Port), "-j", "DROP"); err != nil {
+			return err
+		}
+		if err = ipt.AppendUnique("filter", "INPUT", "-i", config.DockerBridgeName, "-p", "udp", "--dst", dockerBridgeIP.String(), "--dport", fmt.Sprint(config.Port+1), "-j", "DROP"); err != nil {
+			return err
+		}
+
+		// let DNS traffic to weaveDNS, since otherwise it might get blocked by the likes of UFW
+		if err = ipt.AppendUnique("filter", "INPUT", "-i", config.DockerBridgeName, "-p", "udp", "--dport", "53", "-j", "ACCEPT"); err != nil {
+			return err
+		}
+		if err = ipt.AppendUnique("filter", "INPUT", "-i", config.DockerBridgeName, "-p", "tcp", "--dport", "53", "-j", "ACCEPT"); err != nil {
+			return err
+		}
+	}
+
+	if config.NPC {
+		// Steer traffic via the NPC
+		_ = ipt.NewChain("filter", "WEAVE-NPC") // ignore error because it might already exist
+		// If WEAVE-NPC chain doesn't exist then next line will fail
+		if err = ipt.AppendUnique("filter", "FORWARD", "-o", config.WeaveBridgeName, "-j", "WEAVE-NPC"); err != nil {
+			return err
+		}
+		if err = ipt.AppendUnique("filter", "FORWARD", "-o", config.WeaveBridgeName, "-m", "state", "--state", "NEW", "-j", "NFLOG", "--nflog-group", "86"); err != nil {
+			return err
+		}
+		if err = ipt.AppendUnique("filter", "FORWARD", "-o", config.WeaveBridgeName, "-j", "DROP"); err != nil {
+			return err
+		}
+	} else {
+		// Work around the situation where there are no rules allowing traffic
+		// across our bridge. E.g. ufw
+		if err = ipt.AppendUnique("filter", "FORWARD", "-i", config.WeaveBridgeName, "-o", config.WeaveBridgeName, "-j", "ACCEPT"); err != nil {
+			return err
+		}
+	}
+
+	if !config.NPC {
+		// Create a chain for allowing ingress traffic when the bridge is exposed
+		_ = ipt.NewChain("filter", "WEAVE-EXPOSE")
+		if err = ipt.AppendUnique("filter", "FORWARD", "-o", config.WeaveBridgeName, "-j", "WEAVE-EXPOSE"); err != nil {
+			return err
+		}
+	}
+
+	// Forward from weave to the rest of the world
+	if err = ipt.AppendUnique("filter", "FORWARD", "-i", config.WeaveBridgeName, "!", "-o", config.WeaveBridgeName, "-j", "ACCEPT"); err != nil {
+		return err
+	}
+	// and allow replies back
+	if err = ipt.AppendUnique("filter", "FORWARD", "-o", config.WeaveBridgeName, "-m", "conntrack", "--ctstate", "RELATED,ESTABLISHED", "-j", "ACCEPT"); err != nil {
+		return err
+	}
+
+	// create a chain for masquerading
+	if err = ipt.ClearChain("nat", "WEAVE"); err != nil {
+		return errors.Wrap(err, "clearing WEAVE chain")
+	}
+	return ipt.AppendUnique("nat", "POSTROUTING", "-j", "WEAVE")
+}
+
+func linkSetUpByName(linkName string) error {
+	link, err := netlink.LinkByName(linkName)
+	if err != nil {
+		return errors.Wrapf(err, "setting link up on %q", linkName)
+	}
+	return netlink.LinkSetUp(link)
+}

net/expose.go

@@ -0,0 +1,115 @@
+package net
+
+import (
+	"net"
+	"syscall"
+
+	"github.com/coreos/go-iptables/iptables"
+	"github.com/j-keck/arping"
+	"github.com/pkg/errors"
+	"github.com/vishvananda/netlink"
+)
+
+// Expose makes the network accessible from a host by assigning a given IP address
+// to the weave bridge.
+//
+// List of params:
+// * "bridgeName" - a name of the weave bridge.
+// * "ipAddr" - IP addr to be assigned to the bridge.
+// * "removeDefaultRoute" - whether to remove a default route installed by the kernel (used only in the AWSVPC mode).
+// * "npc" - whether is Weave NPC running.
+func Expose(bridgeName string, ipAddr *net.IPNet, removeDefaultRoute, npc bool) error {
+	ipt, err := iptables.New()
+	if err != nil {
+		return errors.Wrap(err, "iptables.New")
+	}
+	cidr := ipAddr.String()
+
+	if err := addBridgeIPAddr(bridgeName, ipAddr, removeDefaultRoute); err != nil {
+		return errors.Wrap(err, "addBridgeIPAddr")
+	}
+
+	if err := exposeNAT(ipt, cidr); err != nil {
+		return errors.Wrap(err, "exposeNAT")
+	}
+
+	if !npc {
+		// Docker 1.13 has changed a default policy of FORWARD chain to DROP
+		// (https://github.com/moby/moby/pull/28257) which makes containers
+		// inaccessible from a remote host when the bridge is exposed.
+		//
+		// The change breaks e.g. the AWSVPC mode. To overcome this we install
+		// an explicit rule for accepting forwarded ingress traffic to an
+		// exposed subnet.
+		if err := ipt.AppendUnique("filter", "WEAVE-EXPOSE", "-d", cidr, "-j", "ACCEPT"); err != nil {
+			return errors.Wrap(err, "ipt.AppendUnique")
+		}
+	}
+
+	return nil
+}
+
+func addBridgeIPAddr(bridgeName string, addr *net.IPNet, removeDefaultRoute bool) error {
+	link, err := netlink.LinkByName(bridgeName)
+	if err != nil {
+		return errors.Wrapf(err, "addBridgeIPAddr finding bridge %q", bridgeName)
+	}
+	err = netlink.AddrAdd(link, &netlink.Addr{IPNet: addr})
+	// The IP addr might have been already set by a concurrent request, just ignore then
+	if err != nil && err != syscall.Errno(syscall.EEXIST) {
+		return errors.Wrapf(err, "adding address %v to %q", addr, bridgeName)
+	}
+
+	// Sending multiple ARP REQUESTs in the case of EEXIST above does not hurt
+	arping.GratuitousArpOverIfaceByName(addr.IP, bridgeName)
+
+	// Remove a default route installed by the kernel. Required by the AWSVPC mode.
+	if removeDefaultRoute {
+		routeFilter := &netlink.Route{
+			LinkIndex: link.Attrs().Index,
+			Dst:       &net.IPNet{IP: addr.IP.Mask(addr.Mask), Mask: addr.Mask},
+			Protocol:  2, // RTPROT_KERNEL
+		}
+		filterMask := netlink.RT_FILTER_OIF | netlink.RT_FILTER_DST | netlink.RT_FILTER_PROTOCOL
+		routes, err := netlink.RouteListFiltered(netlink.FAMILY_V4, routeFilter, filterMask)
+		if err != nil {
+			return errors.Wrapf(err, "failed to get route list for bridge %q", bridgeName)
+		}
+		for _, r := range routes {
+			err = netlink.RouteDel(&r)
+			// Again, there might be a concurrent request for removing routes
+			if err != nil && err != syscall.Errno(syscall.ESRCH) {
+				return errors.Wrapf(err, "failed to delete default route %+v", r)
+			}
+		}
+	}
+	return nil
+}
+
+func exposeNAT(ipt *iptables.IPTables, cidr string) error {
+	if err := addNatRule(ipt, "-s", cidr, "-d", "224.0.0.0/4", "-j", "RETURN"); err != nil {
+		return err
+	}
+	if err := addNatRule(ipt, "-d", cidr, "!", "-s", cidr, "-j", "MASQUERADE"); err != nil {
+		return err
+	}
+	return addNatRule(ipt, "-s", cidr, "!", "-d", cidr, "-j", "MASQUERADE")
+}
+
+func addNatRule(ipt *iptables.IPTables, rulespec ...string) error {
+	// Loop until we get an exit code other than "temporarily unavailable"
+	for {
+		if err := ipt.AppendUnique("nat", "WEAVE", rulespec...); err != nil {
+			if ierr, ok := err.(*iptables.Error); ok {
+				if status, ok := ierr.ExitError.Sys().(syscall.WaitStatus); ok {
+					// (magic exit code 4 found in iptables source code; undocumented)
+					if status.ExitStatus() == 4 {
+						continue
+					}
+				}
+			}
+			return err
+		}
+		return nil
+	}
+}

net/if.go

@@ -82,3 +82,25 @@ func findInterface(ifaceName string) (iface *net.Interface, err error) {
 	}
 	return
 }
+
+func LocalAddresses() ([]*net.IPNet, error) {
+	links, err := netlink.LinkList()
+	if err != nil {
+		return nil, err
+	}
+
+	retval := []*net.IPNet{}
+	for _, link := range links {
+		if _, isBridge := link.(*netlink.Bridge); isBridge || (link.Attrs().Flags&net.FlagLoopback) != 0 {
+			continue
+		}
+		addrs, err := netlink.AddrList(link, netlink.FAMILY_V4)
+		if err != nil {
+			return nil, err
+		}
+		for _, addr := range addrs {
+			retval = append(retval, addr.IPNet)
+		}
+	}
+	return retval, nil
+}

net/ipsec/ipsec.go

@@ -421,10 +421,7 @@ func (ipsec *IPSec) installDropNonEncrypted(srcIP, dstIP net.IP, udpPort int, in
 
 func (ipsec *IPSec) removeDropNonEncrypted(srcIP, dstIP net.IP, udpPort int, inSPI SPI) error {
 	rules := rulesDropNonEncrypted(srcIP, dstIP, udpPort, inSPI)
-	if err := ipsec.resetRules(rules, true); err != nil {
-		return err
-	}
-	return nil
+	return ipsec.resetRules(rules, true)
 }
 
 func (ipsec *IPSec) removeDropNonEncryptedInbound(srcIP, dstIP net.IP, inSPI SPI) error {

net/netns.go

@@ -5,6 +5,7 @@ import (
 	"errors"
 	"fmt"
 	"os/exec"
+	"path/filepath"
 	"runtime"
 
 	"github.com/vishvananda/netlink"
@@ -80,5 +81,9 @@ func WithNetNSByPid(pid int, cmd string, args ...string) ([]byte, error) {
 }
 
 func NSPathByPid(pid int) string {
-	return fmt.Sprintf("/proc/%d/ns/net", pid)
+	return NSPathByPidWithRoot("/", pid)
+}
+
+func NSPathByPidWithRoot(root string, pid int) string {
+	return filepath.Join(root, fmt.Sprintf("/proc/%d/ns/net", pid))
 }

net/uuid.go

@@ -36,8 +36,8 @@ func getSystemUUID(hostRoot string) ([]byte, error) {
 	return append(machineid, uuid...), nil
 }
 
-func getPersistedPeerName(pathName string) (mesh.PeerName, error) {
-	d, err := db.NewBoltDBReadOnly(pathName)
+func getPersistedPeerName(dbPrefix string) (mesh.PeerName, error) {
+	d, err := db.NewBoltDBReadOnly(dbPrefix)
 	if err != nil {
 		return mesh.UnknownPeerName, err
 	}
@@ -56,9 +56,8 @@ func getPersistedPeerName(pathName string) (mesh.PeerName, error) {
 func GetSystemPeerName(dbPrefix, hostRoot string) (string, error) {
 	// Check if we have a persisted name that matches the old-style ID for this host
 	if oldUUID, err := getOldStyleSystemUUID(); err == nil {
-		pathName := dbPrefix + db.FileName
-		if _, err := os.Stat(pathName); err == nil {
-			persistedPeerName, err := getPersistedPeerName(pathName)
+		if _, err := os.Stat(db.Pathname(dbPrefix)); err == nil {
+			persistedPeerName, err := getPersistedPeerName(dbPrefix)
 			if err != nil && !os.IsNotExist(err) {
 				return "", err
 			}

net/veth.go

@@ -4,15 +4,12 @@ import (
 	"fmt"
 	"net"
 	"strings"
-	"syscall"
 	"time"
 
 	"github.com/coreos/go-iptables/iptables"
 	"github.com/j-keck/arping"
 	"github.com/vishvananda/netlink"
 	"github.com/vishvananda/netns"
-
-	"github.com/weaveworks/weave/common/odp"
 )
 
 // create and attach a veth to the Weave bridge
@@ -40,22 +37,17 @@ func CreateAndAttachVeth(name, peerName, bridgeName string, mtu int, keepTXOn bo
 		return nil, fmt.Errorf(format, a...)
 	}
 
-	switch bridgeType := DetectBridgeType(bridgeName, DatapathName); bridgeType {
-	case Bridge, BridgedFastdp:
-		if err := netlink.LinkSetMasterByIndex(veth, bridge.Attrs().Index); err != nil {
-			return cleanup(`unable to set master of %s: %s`, name, err)
+	bridgeType, err := ExistingBridgeType(bridgeName, DatapathName)
+	if err != nil {
+		return cleanup("detect bridge type: %s", err)
+	}
+	if err := bridgeType.attach(veth); err != nil {
+		return cleanup("attaching veth %q to %q: %s", name, bridgeName, err)
+	}
+	if !bridgeType.IsFastdp() && !keepTXOn {
+		if err := EthtoolTXOff(veth.PeerName); err != nil {
+			return cleanup(`unable to set tx off on %q: %s`, peerName, err)
 		}
-		if bridgeType == Bridge && !keepTXOn {
-			if err := EthtoolTXOff(peerName); err != nil {
-				return cleanup(`unable to set tx off on %q: %s`, peerName, err)
-			}
-		}
-	case Fastdp:
-		if err := odp.AddDatapathInterface(bridgeName, name); err != nil {
-			return cleanup(`failed to attach %s to device "%s": %s`, name, bridgeName, err)
-		}
-	default:
-		return cleanup(`invalid bridge configuration`)
 	}
 
 	if init != nil {
@@ -111,8 +103,6 @@ func interfaceExistsInNamespace(netNSPath string, ifName string) bool {
 	return err == nil
 }
 
-// NB: This function can be used only by a process that terminates immediately
-//     after calling the function as it changes netns via WithNetNSLinkUnsafe.
 func AttachContainer(netNSPath, id, ifName, bridgeName string, mtu int, withMulticastRoute bool, cidrs []*net.IPNet, keepTXOn bool, hairpinMode bool) error {
 	ns, err := netns.GetFromPath(netNSPath)
 	if err != nil {
@@ -120,11 +110,6 @@ func AttachContainer(netNSPath, id, ifName, bridgeName string, mtu int, withMult
 	}
 	defer ns.Close()
 
-	ipt, err := iptables.New()
-	if err != nil {
-		return err
-	}
-
 	if !interfaceExistsInNamespace(netNSPath, ifName) {
 		maxIDLen := IFNAMSIZ - 1 - len(vethPrefix+"pl")
 		if len(id) > maxIDLen {
@@ -149,57 +134,92 @@ func AttachContainer(netNSPath, id, ifName, bridgeName string, mtu int, withMult
 
 	}
 
-	if err := WithNetNSLinkUnsafe(ns, ifName, func(veth netlink.Link) error {
-		newAddresses, err := AddAddresses(veth, cidrs)
-		if err != nil {
-			return err
-		}
+	args := []string{ifName, fmt.Sprintf("%t", withMulticastRoute)}
+	for _, cidr := range cidrs {
+		args = append(args, cidr.String())
+	}
+	if _, err := WithNetNS(netNSPath, "setup-iface-addrs", args...); err != nil {
+		return fmt.Errorf("error setting up interface addresses: %s", err)
+	}
+	return nil
+}
 
-		// Add multicast ACCEPT rules for new subnets
-		for _, ipnet := range newAddresses {
-			acceptRule := []string{"-i", ifName, "-s", subnet(ipnet), "-d", "224.0.0.0/4", "-j", "ACCEPT"}
-			exists, err := ipt.Exists("filter", "INPUT", acceptRule...)
-			if err != nil {
-				return err
-			}
-			if !exists {
-				if err := ipt.Insert("filter", "INPUT", 1, acceptRule...); err != nil {
-					return err
-				}
-			}
-		}
-
-		if err := netlink.LinkSetUp(veth); err != nil {
-			return err
-		}
-		for _, ipnet := range newAddresses {
-			// If we don't wait for a bit here, we see the arp fail to reach the bridge.
-			time.Sleep(1 * time.Millisecond)
-			arping.GratuitousArpOverIfaceByName(ipnet.IP, ifName)
-		}
-		if withMulticastRoute {
-			/* Route multicast packets across the weave network.
-			This must come last in 'attach'. If you change this, change weavewait to match.
-
-			TODO: Add the MTU lock to prevent PMTU discovery for multicast
-			destinations. Without that, the kernel sets the DF flag on
-			multicast packets. Since RFC1122 prohibits sending of ICMP
-			errors for packets with multicast destinations, that causes
-			packets larger than the PMTU to be dropped silently.  */
-
-			_, multicast, _ := net.ParseCIDR("224.0.0.0/4")
-			if err := AddRoute(veth, netlink.SCOPE_LINK, multicast, nil); err != nil {
-				return err
-			}
-		}
-		return nil
-	}); err != nil {
+// SetupIfaceAddrs is the implementation of the 'setup-iface-addrs' call above,
+// running in another process in the container's netns
+func SetupIfaceAddrs(veth netlink.Link, withMulticastRoute bool, cidrs []*net.IPNet) error {
+	newAddresses, err := AddAddresses(veth, cidrs)
+	if err != nil {
 		return err
 	}
 
+	ifName := veth.Attrs().Name
+	ipt, err := iptables.New()
+	if err != nil {
+		return err
+	}
+
+	// Add multicast ACCEPT rules for new subnets
+	for _, ipnet := range newAddresses {
+		acceptRule := []string{"-i", ifName, "-s", subnet(ipnet), "-d", "224.0.0.0/4", "-j", "ACCEPT"}
+		exists, err := ipt.Exists("filter", "INPUT", acceptRule...)
+		if err != nil {
+			return err
+		}
+		if !exists {
+			if err := ipt.Insert("filter", "INPUT", 1, acceptRule...); err != nil {
+				return err
+			}
+		}
+	}
+
+	if err := netlink.LinkSetUp(veth); err != nil {
+		return err
+	}
+	for _, ipnet := range newAddresses {
+		// If we don't wait for a bit here, we see the arp fail to reach the bridge.
+		time.Sleep(1 * time.Millisecond)
+		arping.GratuitousArpOverIfaceByName(ipnet.IP, ifName)
+	}
+	if withMulticastRoute {
+		/* Route multicast packets across the weave network.
+		This must come last in 'attach'. If you change this, change weavewait to match.
+
+		TODO: Add the MTU lock to prevent PMTU discovery for multicast
+		destinations. Without that, the kernel sets the DF flag on
+		multicast packets. Since RFC1122 prohibits sending of ICMP
+		errors for packets with multicast destinations, that causes
+		packets larger than the PMTU to be dropped silently.  */
+
+		_, multicast, _ := net.ParseCIDR("224.0.0.0/4")
+		if err := AddRoute(veth, netlink.SCOPE_LINK, multicast, nil); err != nil {
+			return err
+		}
+	}
 	return nil
 }
 
+// SetupIface is the implementation of the 'setup-iface' call above,
+// running in another process in the container's netns
+func SetupIface(ifaceName, newIfName string) error {
+	ipt, err := iptables.New()
+	if err != nil {
+		return err
+	}
+
+	link, err := netlink.LinkByName(ifaceName)
+	if err != nil {
+		return err
+	}
+	if err := netlink.LinkSetName(link, newIfName); err != nil {
+		return err
+	}
+	// This is only called by AttachContainer which is only called in host pid namespace
+	if err := ConfigureARPCache("/proc", newIfName); err != nil {
+		return err
+	}
+	return ipt.Append("filter", "INPUT", "-i", newIfName, "-d", "224.0.0.0/4", "-j", "DROP")
+}
+
 // NB: This function can be used only by a process that terminates immediately
 //     after calling the function as it changes netns via WithNetNSLinkUnsafe.
 func DetachContainer(netNSPath, id, ifName string, cidrs []*net.IPNet) error {
@@ -276,39 +296,3 @@ func subnets(addrs []netlink.Addr) map[string]struct{} {
 	}
 	return subnets
 }
-
-func addNatRule(ipt *iptables.IPTables, rulespec ...string) error {
-	// Loop until we get an exit code other than "temporarily unavailable"
-	for {
-		if err := ipt.AppendUnique("nat", "WEAVE", rulespec...); err != nil {
-			if ierr, ok := err.(*iptables.Error); ok {
-				if status, ok := ierr.ExitError.Sys().(syscall.WaitStatus); ok {
-					// (magic exit code 4 found in iptables source code; undocumented)
-					if status.ExitStatus() == 4 {
-						continue
-					}
-				}
-			}
-			return err
-		}
-		return nil
-	}
-}
-
-func ExposeNAT(ipnet net.IPNet) error {
-	ipt, err := iptables.New()
-	if err != nil {
-		return err
-	}
-	cidr := ipnet.String()
-	if err := addNatRule(ipt, "-s", cidr, "-d", "224.0.0.0/4", "-j", "RETURN"); err != nil {
-		return err
-	}
-	if err := addNatRule(ipt, "-d", cidr, "!", "-s", cidr, "-j", "MASQUERADE"); err != nil {
-		return err
-	}
-	if err := addNatRule(ipt, "-s", cidr, "!", "-d", cidr, "-j", "MASQUERADE"); err != nil {
-		return err
-	}
-	return nil
-}

npc/analyser.go

@@ -3,14 +3,19 @@ package npc
 import (
 	"fmt"
 
-	"k8s.io/client-go/pkg/api"
-	extnapi "k8s.io/client-go/pkg/apis/extensions/v1beta1"
-	"k8s.io/client-go/pkg/util/intstr"
+	apiv1 "k8s.io/api/core/v1"
+	extnapi "k8s.io/api/extensions/v1beta1"
+	networkingv1 "k8s.io/api/networking/v1"
+	"k8s.io/apimachinery/pkg/util/intstr"
 
 	"github.com/weaveworks/weave/npc/ipset"
 )
 
-func (ns *ns) analysePolicy(policy *extnapi.NetworkPolicy) (
+// analysePolicyLegacy is used to analyse extensions/v1beta1/NetworkPolicy (legacy) and
+// implements pre-1.7 k8s netpol semantics, whilst analysePolicy - networking.k8s.io/v1/NetworkPolicy
+// and 1.7 semantics.
+
+func (ns *ns) analysePolicyLegacy(policy *extnapi.NetworkPolicy) (
 	rules map[string]*ruleSpec,
 	nsSelectors, podSelectors map[string]*selectorSpec,
 	err error) {
@@ -19,7 +24,7 @@ func (ns *ns) analysePolicy(policy *extnapi.NetworkPolicy) (
 	podSelectors = make(map[string]*selectorSpec)
 	rules = make(map[string]*ruleSpec)
 
-	dstSelector, err := newSelectorSpec(&policy.Spec.PodSelector, ns.name, ipset.HashIP)
+	dstSelector, err := newSelectorSpec(&policy.Spec.PodSelector, true, ns.name, ipset.HashIP)
 	if err != nil {
 		return nil, nil, nil, err
 	}
@@ -45,7 +50,7 @@ func (ns *ns) analysePolicy(policy *extnapi.NetworkPolicy) (
 			} else {
 				// Ports is present and contains at least one item, then this rule allows traffic
 				// only if the traffic matches at least one port in the ports list.
-				withNormalisedProtoAndPort(ingressRule.Ports, func(proto, port string) {
+				withNormalisedProtoAndPortLegacy(ingressRule.Ports, func(proto, port string) {
 					rule := newRuleSpec(&proto, nil, dstSelector, &port)
 					rules[rule.key] = rule
 				})
@@ -56,14 +61,14 @@ func (ns *ns) analysePolicy(policy *extnapi.NetworkPolicy) (
 			for _, peer := range ingressRule.From {
 				var srcSelector *selectorSpec
 				if peer.PodSelector != nil {
-					srcSelector, err = newSelectorSpec(peer.PodSelector, ns.name, ipset.HashIP)
+					srcSelector, err = newSelectorSpec(peer.PodSelector, false, ns.name, ipset.HashIP)
 					if err != nil {
 						return nil, nil, nil, err
 					}
 					podSelectors[srcSelector.key] = srcSelector
 				}
 				if peer.NamespaceSelector != nil {
-					srcSelector, err = newSelectorSpec(peer.NamespaceSelector, "", ipset.ListSet)
+					srcSelector, err = newSelectorSpec(peer.NamespaceSelector, false, "", ipset.ListSet)
 					if err != nil {
 						return nil, nil, nil, err
 					}
@@ -77,6 +82,80 @@ func (ns *ns) analysePolicy(policy *extnapi.NetworkPolicy) (
 				} else {
 					// Ports is present and contains at least one item, then this rule allows traffic
 					// only if the traffic matches at least one port in the ports list.
+					withNormalisedProtoAndPortLegacy(ingressRule.Ports, func(proto, port string) {
+						rule := newRuleSpec(&proto, srcSelector, dstSelector, &port)
+						rules[rule.key] = rule
+					})
+				}
+			}
+		}
+	}
+
+	return rules, nsSelectors, podSelectors, nil
+}
+
+func (ns *ns) analysePolicy(policy *networkingv1.NetworkPolicy) (
+	rules map[string]*ruleSpec,
+	nsSelectors, podSelectors map[string]*selectorSpec,
+	err error) {
+
+	nsSelectors = make(map[string]*selectorSpec)
+	podSelectors = make(map[string]*selectorSpec)
+	rules = make(map[string]*ruleSpec)
+
+	// If empty, matches all pods in a namespace
+	dstSelector, err := newSelectorSpec(&policy.Spec.PodSelector, true, ns.name, ipset.HashIP)
+	if err != nil {
+		return nil, nil, nil, err
+	}
+	podSelectors[dstSelector.key] = dstSelector
+
+	// If ingress is empty then this NetworkPolicy does not allow any traffic
+	if policy.Spec.Ingress == nil || len(policy.Spec.Ingress) == 0 {
+		return
+	}
+
+	for _, ingressRule := range policy.Spec.Ingress {
+		// If Ports is empty or missing, this rule matches all ports
+		allPorts := ingressRule.Ports == nil || len(ingressRule.Ports) == 0
+		// If From is empty or missing, this rule matches all sources
+		allSources := ingressRule.From == nil || len(ingressRule.From) == 0
+
+		if allSources {
+			if allPorts {
+				rule := newRuleSpec(nil, nil, dstSelector, nil)
+				rules[rule.key] = rule
+			} else {
+				withNormalisedProtoAndPort(ingressRule.Ports, func(proto, port string) {
+					rule := newRuleSpec(&proto, nil, dstSelector, &port)
+					rules[rule.key] = rule
+				})
+			}
+		} else {
+			for _, peer := range ingressRule.From {
+				var srcSelector *selectorSpec
+
+				// NetworkPolicyPeer describes a peer to allow traffic from.
+				// Exactly one of its fields must be specified.
+				if peer.PodSelector != nil {
+					srcSelector, err = newSelectorSpec(peer.PodSelector, false, ns.name, ipset.HashIP)
+					if err != nil {
+						return nil, nil, nil, err
+					}
+					podSelectors[srcSelector.key] = srcSelector
+
+				} else if peer.NamespaceSelector != nil {
+					srcSelector, err = newSelectorSpec(peer.NamespaceSelector, false, "", ipset.ListSet)
+					if err != nil {
+						return nil, nil, nil, err
+					}
+					nsSelectors[srcSelector.key] = srcSelector
+				}
+
+				if allPorts {
+					rule := newRuleSpec(nil, srcSelector, dstSelector, nil)
+					rules[rule.key] = rule
+				} else {
 					withNormalisedProtoAndPort(ingressRule.Ports, func(proto, port string) {
 						rule := newRuleSpec(&proto, srcSelector, dstSelector, &port)
 						rules[rule.key] = rule
@@ -89,26 +168,40 @@ func (ns *ns) analysePolicy(policy *extnapi.NetworkPolicy) (
 	return rules, nsSelectors, podSelectors, nil
 }
 
-func withNormalisedProtoAndPort(npps []extnapi.NetworkPolicyPort, f func(proto, port string)) {
+func withNormalisedProtoAndPortLegacy(npps []extnapi.NetworkPolicyPort, f func(proto, port string)) {
 	for _, npp := range npps {
-		// If no proto is specified, default to TCP
-		proto := string(api.ProtocolTCP)
-		if npp.Protocol != nil {
-			proto = string(*npp.Protocol)
-		}
-
-		// If no port is specified, match any port. Let iptables executable handle
-		// service name resolution
-		port := "0:65535"
-		if npp.Port != nil {
-			switch npp.Port.Type {
-			case intstr.Int:
-				port = fmt.Sprintf("%d", npp.Port.IntVal)
-			case intstr.String:
-				port = npp.Port.StrVal
-			}
-		}
-
-		f(proto, port)
+		f(proto(npp.Protocol), port(npp.Port))
 	}
 }
+
+func withNormalisedProtoAndPort(npps []networkingv1.NetworkPolicyPort, f func(proto, port string)) {
+	for _, npp := range npps {
+		f(proto(npp.Protocol), port(npp.Port))
+	}
+}
+
+func proto(p *apiv1.Protocol) string {
+	// If no proto is specified, default to TCP
+	proto := string(apiv1.ProtocolTCP)
+	if p != nil {
+		proto = string(*p)
+	}
+
+	return proto
+}
+
+func port(p *intstr.IntOrString) string {
+	// If no port is specified, match any port. Let iptables executable handle
+	// service name resolution
+	port := "0:65535"
+	if p != nil {
+		switch p.Type {
+		case intstr.Int:
+			port = fmt.Sprintf("%d", p.IntVal)
+		case intstr.String:
+			port = p.StrVal
+		}
+	}
+
+	return port
+}

npc/controller.go

@@ -3,13 +3,14 @@ package npc
 import (
 	"sync"
 
-	"github.com/coreos/go-iptables/iptables"
 	"github.com/pkg/errors"
-	coreapi "k8s.io/client-go/pkg/api/v1"
-	extnapi "k8s.io/client-go/pkg/apis/extensions/v1beta1"
+	coreapi "k8s.io/api/core/v1"
+	extnapi "k8s.io/api/extensions/v1beta1"
+	networkingv1 "k8s.io/api/networking/v1"
 
 	"github.com/weaveworks/weave/common"
 	"github.com/weaveworks/weave/npc/ipset"
+	"github.com/weaveworks/weave/npc/iptables"
 )
 
 type NetworkPolicyController interface {
@@ -21,9 +22,9 @@ type NetworkPolicyController interface {
 	UpdatePod(oldObj, newObj *coreapi.Pod) error
 	DeletePod(obj *coreapi.Pod) error
 
-	AddNetworkPolicy(obj *extnapi.NetworkPolicy) error
-	UpdateNetworkPolicy(oldObj, newObj *extnapi.NetworkPolicy) error
-	DeleteNetworkPolicy(obj *extnapi.NetworkPolicy) error
+	AddNetworkPolicy(obj interface{}) error
+	UpdateNetworkPolicy(oldObj, newObj interface{}) error
+	DeleteNetworkPolicy(obj interface{}) error
 }
 
 type controller struct {
@@ -31,21 +32,25 @@ type controller struct {
 
 	nodeName string // my node name
 
-	ipt *iptables.IPTables
+	ipt iptables.Interface
 	ips ipset.Interface
 
 	nss         map[string]*ns // ns name -> ns struct
 	nsSelectors *selectorSet   // selector string -> nsSelector
+
+	legacy bool // use legacy network policy semantics (k8s pre-1.7)
 }
 
-func New(nodeName string, ipt *iptables.IPTables, ips ipset.Interface) NetworkPolicyController {
+func New(nodeName string, legacy bool, ipt iptables.Interface, ips ipset.Interface) NetworkPolicyController {
 	c := &controller{
 		nodeName: nodeName,
+		legacy:   legacy,
 		ipt:      ipt,
 		ips:      ips,
 		nss:      make(map[string]*ns)}
 
-	c.nsSelectors = newSelectorSet(ips, c.onNewNsSelector)
+	doNothing := func(*selector) error { return nil }
+	c.nsSelectors = newSelectorSet(ips, c.onNewNsSelector, doNothing, doNothing)
 
 	return c
 }
@@ -54,7 +59,7 @@ func (npc *controller) onNewNsSelector(selector *selector) error {
 	for _, ns := range npc.nss {
 		if ns.namespace != nil {
 			if selector.matches(ns.namespace.ObjectMeta.Labels) {
-				if err := ns.ips.AddEntry(selector.spec.ipsetName, string(ns.allPods.ipsetName)); err != nil {
+				if err := selector.addEntry(ns.namespace.ObjectMeta.UID, string(ns.allPods.ipsetName), namespaceComment(ns)); err != nil {
 					return err
 				}
 			}
@@ -66,7 +71,7 @@ func (npc *controller) onNewNsSelector(selector *selector) error {
 func (npc *controller) withNS(name string, f func(ns *ns) error) error {
 	ns, found := npc.nss[name]
 	if !found {
-		newNs, err := newNS(name, npc.ipt, npc.ips, npc.nsSelectors)
+		newNs, err := newNS(name, npc.nodeName, npc.legacy, npc.ipt, npc.ips, npc.nsSelectors)
 		if err != nil {
 			return err
 		}
@@ -85,21 +90,10 @@ func (npc *controller) withNS(name string, f func(ns *ns) error) error {
 	return nil
 }
 
-func (npc *controller) checkLocalPod(obj *coreapi.Pod, event string) bool {
-	if obj.Spec.NodeName != npc.nodeName {
-		common.Log.Debugf("%s ignored for pod %s/%s on node %s", event, obj.ObjectMeta.Namespace, obj.ObjectMeta.Name, obj.Spec.NodeName)
-		return false
-	}
-	return true
-}
-
 func (npc *controller) AddPod(obj *coreapi.Pod) error {
 	npc.Lock()
 	defer npc.Unlock()
 
-	if !npc.checkLocalPod(obj, "AddPod") {
-		return nil
-	}
 	common.Log.Debugf("EVENT AddPod %s", js(obj))
 	return npc.withNS(obj.ObjectMeta.Namespace, func(ns *ns) error {
 		return errors.Wrap(ns.addPod(obj), "add pod")
@@ -110,10 +104,6 @@ func (npc *controller) UpdatePod(oldObj, newObj *coreapi.Pod) error {
 	npc.Lock()
 	defer npc.Unlock()
 
-	// Pods don't move across nodes so we only check the new version
-	if !npc.checkLocalPod(newObj, "UpdatePod") {
-		return nil
-	}
 	common.Log.Debugf("EVENT UpdatePod %s %s", js(oldObj), js(newObj))
 	return npc.withNS(oldObj.ObjectMeta.Namespace, func(ns *ns) error {
 		return errors.Wrap(ns.updatePod(oldObj, newObj), "update pod")
@@ -124,41 +114,53 @@ func (npc *controller) DeletePod(obj *coreapi.Pod) error {
 	npc.Lock()
 	defer npc.Unlock()
 
-	if !npc.checkLocalPod(obj, "DeletePod") {
-		return nil
-	}
 	common.Log.Debugf("EVENT DeletePod %s", js(obj))
 	return npc.withNS(obj.ObjectMeta.Namespace, func(ns *ns) error {
 		return errors.Wrap(ns.deletePod(obj), "delete pod")
 	})
 }
 
-func (npc *controller) AddNetworkPolicy(obj *extnapi.NetworkPolicy) error {
+func (npc *controller) AddNetworkPolicy(obj interface{}) error {
 	npc.Lock()
 	defer npc.Unlock()
 
+	nsName, err := nsName(obj)
+	if err != nil {
+		return err
+	}
+
 	common.Log.Infof("EVENT AddNetworkPolicy %s", js(obj))
-	return npc.withNS(obj.ObjectMeta.Namespace, func(ns *ns) error {
+	return npc.withNS(nsName, func(ns *ns) error {
 		return errors.Wrap(ns.addNetworkPolicy(obj), "add network policy")
 	})
 }
 
-func (npc *controller) UpdateNetworkPolicy(oldObj, newObj *extnapi.NetworkPolicy) error {
+func (npc *controller) UpdateNetworkPolicy(oldObj, newObj interface{}) error {
 	npc.Lock()
 	defer npc.Unlock()
 
+	nsName, err := nsName(oldObj)
+	if err != nil {
+		return err
+	}
+
 	common.Log.Infof("EVENT UpdateNetworkPolicy %s %s", js(oldObj), js(newObj))
-	return npc.withNS(oldObj.ObjectMeta.Namespace, func(ns *ns) error {
+	return npc.withNS(nsName, func(ns *ns) error {
 		return errors.Wrap(ns.updateNetworkPolicy(oldObj, newObj), "update network policy")
 	})
 }
 
-func (npc *controller) DeleteNetworkPolicy(obj *extnapi.NetworkPolicy) error {
+func (npc *controller) DeleteNetworkPolicy(obj interface{}) error {
 	npc.Lock()
 	defer npc.Unlock()
 
+	nsName, err := nsName(obj)
+	if err != nil {
+		return err
+	}
+
 	common.Log.Infof("EVENT DeleteNetworkPolicy %s", js(obj))
-	return npc.withNS(obj.ObjectMeta.Namespace, func(ns *ns) error {
+	return npc.withNS(nsName, func(ns *ns) error {
 		return errors.Wrap(ns.deleteNetworkPolicy(obj), "delete network policy")
 	})
 }
@@ -192,3 +194,14 @@ func (npc *controller) DeleteNamespace(obj *coreapi.Namespace) error {
 		return errors.Wrap(ns.deleteNamespace(obj), "delete namespace")
 	})
 }
+
+func nsName(obj interface{}) (string, error) {
+	switch obj := obj.(type) {
+	case *networkingv1.NetworkPolicy:
+		return obj.ObjectMeta.Namespace, nil
+	case *extnapi.NetworkPolicy:
+		return obj.ObjectMeta.Namespace, nil
+	}
+
+	return "", errInvalidNetworkPolicyObjType
+}

npc/controller_test.go

@@ -0,0 +1,378 @@
+package npc
+
+import (
+	"log"
+	"testing"
+
+	"github.com/pkg/errors"
+	"github.com/stretchr/testify/require"
+	"github.com/weaveworks/weave/npc/ipset"
+	coreapi "k8s.io/api/core/v1"
+	networkingv1 "k8s.io/api/networking/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/types"
+	"k8s.io/apimachinery/pkg/util/intstr"
+)
+
+type mockSet struct {
+	name    ipset.Name
+	setType ipset.Type
+	subSets map[string]map[types.UID]bool
+}
+
+type mockIPSet struct {
+	sets map[string]mockSet
+}
+
+func newMockIPSet() mockIPSet {
+	return mockIPSet{sets: make(map[string]mockSet)}
+}
+
+func (i *mockIPSet) Create(ipsetName ipset.Name, ipsetType ipset.Type) error {
+	if _, ok := i.sets[string(ipsetName)]; ok {
+		return errors.Errorf("ipset %s already exists", ipsetName)
+	}
+	i.sets[string(ipsetName)] = mockSet{name: ipsetName, setType: ipsetType, subSets: make(map[string]map[types.UID]bool)}
+	return nil
+}
+
+func (i *mockIPSet) AddEntry(user types.UID, ipsetName ipset.Name, entry string, comment string) error {
+	log.Printf("adding entry %s to %s for %s", entry, ipsetName, user)
+	if _, ok := i.sets[string(ipsetName)]; !ok {
+		return errors.Errorf("%s does not exist", entry)
+	}
+	if i.sets[string(ipsetName)].subSets[entry] == nil {
+		i.sets[string(ipsetName)].subSets[entry] = make(map[types.UID]bool)
+	}
+	if _, ok := i.sets[string(ipsetName)].subSets[entry][user]; ok {
+		return errors.Errorf("user %s already owns entry %s", user, entry)
+	}
+	i.sets[string(ipsetName)].subSets[entry][user] = true
+
+	return nil
+}
+
+func (i *mockIPSet) DelEntry(user types.UID, ipsetName ipset.Name, entry string) error {
+	log.Printf("deleting entry %s from %s for %s", entry, ipsetName, user)
+	if _, ok := i.sets[string(ipsetName)]; !ok {
+		return errors.Errorf("ipset %s does not exist", ipsetName)
+	}
+	if _, ok := i.sets[string(ipsetName)].subSets[entry][user]; !ok {
+		return errors.Errorf("user %s does not own entry %s", user, entry)
+	}
+	delete(i.sets[string(ipsetName)].subSets[entry], user)
+
+	if len(i.sets[string(ipsetName)].subSets[entry]) == 0 {
+		delete(i.sets[string(ipsetName)].subSets, entry)
+	}
+
+	return nil
+}
+
+func (i *mockIPSet) Exist(user types.UID, ipsetName ipset.Name, entry string) bool {
+	_, ok := i.sets[string(ipsetName)].subSets[entry][user]
+	return ok
+}
+
+func (i *mockIPSet) entryExists(ipsetName ipset.Name, entry string) bool {
+	return len(i.sets[string(ipsetName)].subSets[entry]) > 0
+}
+
+func (i *mockIPSet) Flush(ipsetName ipset.Name) error {
+	return errors.New("Not Implemented")
+}
+
+func (i *mockIPSet) FlushAll() error {
+	return errors.New("Not Implemented")
+}
+
+func (i *mockIPSet) Destroy(ipsetName ipset.Name) error {
+	if _, ok := i.sets[string(ipsetName)]; !ok {
+		return errors.Errorf("ipset %s does not exist", ipsetName)
+	}
+	delete(i.sets, string(ipsetName))
+	return nil
+}
+
+func (i *mockIPSet) DestroyAll() error {
+	return errors.New("Not Implemented")
+}
+
+func (i *mockIPSet) List(prefix string) ([]ipset.Name, error) {
+	return []ipset.Name{}, errors.New("Not Implemented")
+}
+
+type mockIPTables struct {
+}
+
+func (ipt *mockIPTables) Append(table, chain string, rulespec ...string) error {
+	return nil
+}
+
+func (ipt *mockIPTables) Delete(table, chain string, rulespec ...string) error {
+	return nil
+}
+
+func (ipt *mockIPTables) Insert(table, chain string, pos int, rulespec ...string) error {
+	return nil
+}
+
+func TestRegressionPolicyNamespaceOrdering3059(t *testing.T) {
+	// Test for race condition between namespace and networkpolicy events
+	// https://github.com/weaveworks/weave/issues/3059
+
+	sourceNamespace := &coreapi.Namespace{
+		ObjectMeta: metav1.ObjectMeta{
+			Name: "source",
+			Labels: map[string]string{
+				"app": "source",
+			},
+		},
+	}
+
+	destinationNamespace := &coreapi.Namespace{
+		ObjectMeta: metav1.ObjectMeta{
+			Name: "destination",
+		},
+	}
+
+	port := intstr.FromInt(12345)
+
+	networkPolicy := &networkingv1.NetworkPolicy{
+		ObjectMeta: metav1.ObjectMeta{
+			Name:      "network-policy",
+			Namespace: "destination",
+		},
+		Spec: networkingv1.NetworkPolicySpec{
+			Ingress: []networkingv1.NetworkPolicyIngressRule{
+				{
+					From: []networkingv1.NetworkPolicyPeer{
+						{
+							NamespaceSelector: &metav1.LabelSelector{
+								MatchLabels: map[string]string{
+									"app": "source",
+								},
+							},
+						},
+					},
+					Ports: []networkingv1.NetworkPolicyPort{
+						{
+							Port: &port,
+						},
+					},
+				},
+			},
+		},
+	}
+
+	// Namespaces first
+	m := newMockIPSet()
+	controller := New("foo", false, &mockIPTables{}, &m)
+
+	const (
+		selectorIPSetName = "weave-I239Zp%sCvoVt*D6u=A!2]YEk"
+		sourceIPSetName   = "weave-HboJG1fGgG]/SR%k9H#hv5e96"
+	)
+
+	controller.AddNamespace(sourceNamespace)
+	controller.AddNamespace(destinationNamespace)
+
+	controller.AddNetworkPolicy(networkPolicy)
+
+	require.Equal(t, true, len(m.sets[selectorIPSetName].subSets[sourceIPSetName]) > 0)
+
+	// NetworkPolicy first
+	m = newMockIPSet()
+	controller = New("foo", false, &mockIPTables{}, &m)
+
+	controller.AddNetworkPolicy(networkPolicy)
+
+	controller.AddNamespace(sourceNamespace)
+	controller.AddNamespace(destinationNamespace)
+
+	require.Equal(t, true, len(m.sets[selectorIPSetName].subSets[sourceIPSetName]) > 0)
+}
+
+// Tests default-allow ipset behavior when running in non-legacy mode.
+func TestDefaultAllow(t *testing.T) {
+	const (
+		defaultAllowIPSetName = "weave-E.1.0W^NGSp]0_t5WwH/]gX@L"
+		fooPodIP              = "10.32.0.10"
+		barPodIP              = "10.32.0.11"
+		barPodNewIP           = "10.32.0.12"
+	)
+
+	m := newMockIPSet()
+	controller := New("bar", false, &mockIPTables{}, &m)
+
+	defaultNamespace := &coreapi.Namespace{
+		ObjectMeta: metav1.ObjectMeta{
+			Name: "default",
+		},
+	}
+	controller.AddNamespace(defaultNamespace)
+
+	// Should create an ipset for default-allow
+	require.Contains(t, m.sets, defaultAllowIPSetName)
+
+	podFoo := &coreapi.Pod{
+		ObjectMeta: metav1.ObjectMeta{
+			UID:       "foo",
+			Namespace: "default",
+			Name:      "foo",
+			Labels:    map[string]string{"run": "foo"}},
+		Status: coreapi.PodStatus{PodIP: fooPodIP}}
+	controller.AddPod(podFoo)
+
+	// Should add the foo pod to default-allow
+	require.True(t, m.entryExists(defaultAllowIPSetName, fooPodIP))
+
+	podBar := &coreapi.Pod{
+		ObjectMeta: metav1.ObjectMeta{
+			UID:       "bar",
+			Namespace: "default",
+			Name:      "bar",
+			Labels:    map[string]string{"run": "bar"}},
+		Status: coreapi.PodStatus{PodIP: barPodIP}}
+	podBarNoIP := &coreapi.Pod{ObjectMeta: podBar.ObjectMeta}
+	controller.AddPod(podBarNoIP)
+	controller.UpdatePod(podBarNoIP, podBar)
+
+	// Should add the bar pod to default-allow
+	require.True(t, m.entryExists(defaultAllowIPSetName, barPodIP))
+
+	// Allow access from the bar pod to the foo pod
+	netpol := &networkingv1.NetworkPolicy{
+		ObjectMeta: metav1.ObjectMeta{
+			Name:      "allow-from-bar-to-foo",
+			Namespace: "default",
+		},
+		Spec: networkingv1.NetworkPolicySpec{
+			PodSelector: metav1.LabelSelector{MatchLabels: map[string]string{"run": "foo"}},
+			Ingress: []networkingv1.NetworkPolicyIngressRule{{
+				From: []networkingv1.NetworkPolicyPeer{{
+					PodSelector: &metav1.LabelSelector{
+						MatchLabels: map[string]string{"run": "bar"},
+					},
+				}},
+			}},
+		},
+	}
+	controller.AddNetworkPolicy(netpol)
+
+	// Should remove the foo pod from default-allow as the netpol selects it
+	require.False(t, m.entryExists(defaultAllowIPSetName, fooPodIP))
+	require.True(t, m.entryExists(defaultAllowIPSetName, barPodIP))
+
+	podBarWithNewIP := *podBar
+	podBarWithNewIP.Status.PodIP = barPodNewIP
+	controller.UpdatePod(podBar, &podBarWithNewIP)
+
+	// Should update IP addr of the bar pod in default-allow
+	require.False(t, m.entryExists(defaultAllowIPSetName, barPodIP))
+	require.True(t, m.entryExists(defaultAllowIPSetName, barPodNewIP))
+
+	controller.UpdatePod(&podBarWithNewIP, podBarNoIP)
+	// Should remove the bar pod from default-allow as it does not have any IP addr
+	require.False(t, m.entryExists(defaultAllowIPSetName, barPodNewIP))
+
+	podFooWithNewLabel := *podFoo
+	podFooWithNewLabel.ObjectMeta.Labels = map[string]string{"run": "new-foo"}
+	controller.UpdatePod(podFoo, &podFooWithNewLabel)
+
+	// Should bring back the foo pod to default-allow as it does not match dst of any netpol
+	require.True(t, m.entryExists(defaultAllowIPSetName, fooPodIP))
+
+	controller.UpdatePod(&podFooWithNewLabel, podFoo)
+	// Should remove from default-allow as it matches the netpol after the update
+	require.False(t, m.entryExists(defaultAllowIPSetName, fooPodIP))
+
+	controller.DeleteNetworkPolicy(netpol)
+	// Should bring back the foo pod to default-allow as no netpol selects it
+	require.True(t, m.entryExists(defaultAllowIPSetName, fooPodIP))
+
+	controller.DeletePod(podFoo)
+	// Should remove foo pod from default-allow
+	require.False(t, m.entryExists(defaultAllowIPSetName, fooPodIP))
+}
+
+func TestOutOfOrderPodEvents(t *testing.T) {
+	const (
+		defaultAllowIPSetName = "weave-E.1.0W^NGSp]0_t5WwH/]gX@L"
+		runBarIPSetName       = "weave-bZ~x=yBgzH)Ht()K*Uv3z{M]Y"
+		podIP                 = "10.32.0.10"
+	)
+
+	m := newMockIPSet()
+	controller := New("qux", false, &mockIPTables{}, &m)
+
+	defaultNamespace := &coreapi.Namespace{
+		ObjectMeta: metav1.ObjectMeta{
+			Name: "default",
+		},
+	}
+	controller.AddNamespace(defaultNamespace)
+
+	podFoo := &coreapi.Pod{
+		ObjectMeta: metav1.ObjectMeta{
+			UID:       "foo",
+			Namespace: "default",
+			Name:      "foo",
+			Labels:    map[string]string{"run": "foo"}},
+		Status: coreapi.PodStatus{PodIP: podIP}}
+	controller.AddPod(podFoo)
+
+	// Should be in default-allow as no netpol selects podFoo
+	require.True(t, m.entryExists(defaultAllowIPSetName, podIP))
+
+	netpol := &networkingv1.NetworkPolicy{
+		ObjectMeta: metav1.ObjectMeta{
+			Name:      "allow-from-bar-to-foo",
+			Namespace: "default",
+		},
+		Spec: networkingv1.NetworkPolicySpec{
+			PodSelector: metav1.LabelSelector{MatchLabels: map[string]string{"run": "foo"}},
+			Ingress: []networkingv1.NetworkPolicyIngressRule{{
+				From: []networkingv1.NetworkPolicyPeer{{
+					PodSelector: &metav1.LabelSelector{
+						MatchLabels: map[string]string{"run": "bar"},
+					},
+				}},
+			}},
+		},
+	}
+	controller.AddNetworkPolicy(netpol)
+
+	// Shouldn't be in default-allow as netpol above selects podFoo
+	require.False(t, m.entryExists(defaultAllowIPSetName, podIP))
+
+	podBar := &coreapi.Pod{
+		ObjectMeta: metav1.ObjectMeta{
+			UID:       "bar",
+			Namespace: "default",
+			Name:      "bar",
+			Labels:    map[string]string{"run": "bar"}},
+		Status: coreapi.PodStatus{PodIP: podIP}}
+	controller.AddPod(podBar)
+
+	// Should be in default-allow as no netpol selects podBar
+	require.True(t, m.entryExists(defaultAllowIPSetName, podIP))
+	require.True(t, m.Exist(podBar.ObjectMeta.UID, defaultAllowIPSetName, podIP))
+	// Should be in run=bar ipset
+	require.True(t, m.entryExists(runBarIPSetName, podIP))
+
+	controller.DeletePod(podFoo)
+	// Multiple duplicate events should not affect npc state
+	controller.DeletePod(podFoo)
+	controller.DeletePod(podFoo)
+
+	// Should be in default-allow as no netpol selects podBar and podFoo removal
+	// should not affect podBar in default-allow
+	require.True(t, m.entryExists(defaultAllowIPSetName, podIP))
+
+	controller.DeletePod(podBar)
+
+	// Should remove from default-allow and run=bar ipsets
+	require.Equal(t, 0, len(m.sets[defaultAllowIPSetName].subSets))
+	require.False(t, m.entryExists(runBarIPSetName, podIP))
+}

npc/ipset/ipset.go

@@ -6,6 +6,7 @@ import (
 	"strings"
 
 	"github.com/pkg/errors"
+	"k8s.io/apimachinery/pkg/types"
 )
 
 type Name string
@@ -19,8 +20,9 @@ const (
 
 type Interface interface {
 	Create(ipsetName Name, ipsetType Type) error
-	AddEntry(ipsetName Name, entry string) error
-	DelEntry(ipsetName Name, entry string) error
+	AddEntry(user types.UID, ipsetName Name, entry string, comment string) error
+	DelEntry(user types.UID, ipsetName Name, entry string) error
+	Exist(user types.UID, ipsetName Name, entry string) bool
 	Flush(ipsetName Name) error
 	Destroy(ipsetName Name) error
 
@@ -30,52 +32,100 @@ type Interface interface {
 	DestroyAll() error
 }
 
+type entryKey struct {
+	ipsetName Name
+	entry     string
+}
+
 type ipset struct {
-	refCount
 	*log.Logger
+	enableComments bool
+	// List of users per ipset entry. User is either a namespace or a pod.
+	// There might be multiple users for the same ipset & entry pair because
+	// events from k8s API server might be out of order causing duplicate IPs:
+	// https://github.com/weaveworks/weave/issues/2792.
+	users map[entryKey]map[types.UID]struct{}
 }
 
 func New(logger *log.Logger) Interface {
-	return &ipset{refCount: newRefCount(), Logger: logger}
+	ips := &ipset{
+		Logger:         logger,
+		enableComments: true,
+		users:          make(map[entryKey]map[types.UID]struct{}),
+	}
+
+	// Check for comment support
+	testIpsetName := Name("weave-test-comment")
+	// Clear it out if it already exists
+	_ = ips.Destroy(testIpsetName)
+	// Test for comment support
+	if err := ips.Create(testIpsetName, HashIP); err != nil {
+		ips.Logger.Printf("failed to create %s; disabling comment support", testIpsetName)
+		ips.enableComments = false
+	}
+	// If it was created, destroy it
+	_ = ips.Destroy(testIpsetName)
+
+	return ips
 }
 
 func (i *ipset) Create(ipsetName Name, ipsetType Type) error {
-	return doExec("create", string(ipsetName), string(ipsetType))
+	args := []string{"create", string(ipsetName), string(ipsetType)}
+	if i.enableComments {
+		args = append(args, "comment")
+	}
+	return doExec(args...)
 }
 
-func (i *ipset) AddEntry(ipsetName Name, entry string) error {
-	i.Logger.Printf("adding entry %s to %s", entry, ipsetName)
-	if i.inc(ipsetName, entry) > 1 { // already in the set
+func (i *ipset) AddEntry(user types.UID, ipsetName Name, entry string, comment string) error {
+	i.Logger.Printf("adding entry %s to %s of %s", entry, ipsetName, user)
+
+	if !i.addUser(user, ipsetName, entry) { // already in the set
 		return nil
 	}
-	return doExec("add", string(ipsetName), entry)
+
+	i.Logger.Printf("added entry %s to %s of %s", entry, ipsetName, user)
+
+	args := []string{"add", string(ipsetName), entry}
+	if i.enableComments {
+		args = append(args, "comment", comment)
+	}
+	return doExec(args...)
 }
 
-func (i *ipset) DelEntry(ipsetName Name, entry string) error {
-	i.Logger.Printf("deleting entry %s from %s", entry, ipsetName)
-	if i.dec(ipsetName, entry) > 0 { // still needed
+func (i *ipset) DelEntry(user types.UID, ipsetName Name, entry string) error {
+	i.Logger.Printf("deleting entry %s from %s of %s", entry, ipsetName, user)
+
+	if !i.delUser(user, ipsetName, entry) { // still needed
 		return nil
 	}
+
+	i.Logger.Printf("deleted entry %s from %s of %s", entry, ipsetName, user)
+
 	return doExec("del", string(ipsetName), entry)
 }
 
+func (i *ipset) Exist(user types.UID, ipsetName Name, entry string) bool {
+	return i.existUser(user, ipsetName, entry)
+}
+
 func (i *ipset) Flush(ipsetName Name) error {
-	i.removeSet(ipsetName)
+	i.removeSetFromUsers(ipsetName)
 	return doExec("flush", string(ipsetName))
 }
 
 func (i *ipset) FlushAll() error {
-	i.refCount = newRefCount()
+	i.users = make(map[entryKey]map[types.UID]struct{})
 	return doExec("flush")
 }
 
 func (i *ipset) Destroy(ipsetName Name) error {
-	i.removeSet(ipsetName)
+	i.removeSetFromUsers(ipsetName)
 	return doExec("destroy", string(ipsetName))
 }
 
 func (i *ipset) DestroyAll() error {
-	i.refCount = newRefCount()
+	i.users = make(map[entryKey]map[types.UID]struct{})
 	return doExec("destroy")
 }
 
@@ -97,44 +147,52 @@ func (i *ipset) List(prefix string) ([]Name, error) {
 	return selected, err
 }
 
+// Returns true if entry does not exist in ipset (entry has to be inserted into ipset).
+func (i *ipset) addUser(user types.UID, ipsetName Name, entry string) bool {
+	k := entryKey{ipsetName, entry}
+	add := false
+
+	if i.users[k] == nil {
+		i.users[k] = make(map[types.UID]struct{})
+	}
+	if len(i.users[k]) == 0 {
+		add = true
+	}
+	i.users[k][user] = struct{}{}
+
+	return add
+}
+
+// Returns true if user is the last owner of entry (entry has to be removed from ipset).
+func (i *ipset) delUser(user types.UID, ipsetName Name, entry string) bool {
+	k := entryKey{ipsetName, entry}
+
+	oneLeft := len(i.users[k]) == 1
+	delete(i.users[k], user)
+
+	if len(i.users[k]) == 0 {
+		delete(i.users, k)
+	}
+
+	return oneLeft && (len(i.users[k]) == 0)
+}
+
+func (i *ipset) existUser(user types.UID, ipsetName Name, entry string) bool {
+	_, ok := i.users[entryKey{ipsetName, entry}][user]
+	return ok
+}
+
+func (i *ipset) removeSetFromUsers(ipsetName Name) {
+	for k := range i.users {
+		if k.ipsetName == ipsetName {
+			delete(i.users, k)
+		}
+	}
+}
+
 func doExec(args ...string) error {
 	if output, err := exec.Command("ipset", args...).CombinedOutput(); err != nil {
 		return errors.Wrapf(err, "ipset %v failed: %s", args, output)
 	}
 	return nil
 }
-
-// Reference-counting
-type key struct {
-	ipsetName Name
-	entry     string
-}
-
-// note no locking is required as all operations are serialised in the controller
-type refCount struct {
-	ref map[key]int
-}
-
-func newRefCount() refCount {
-	return refCount{ref: make(map[key]int)}
-}
-
-func (rc *refCount) inc(ipsetName Name, entry string) int {
-	k := key{ipsetName, entry}
-	rc.ref[k]++
-	return rc.ref[k]
-}
-
-func (rc *refCount) dec(ipsetName Name, entry string) int {
-	k := key{ipsetName, entry}
-	rc.ref[k]--
-	return rc.ref[k]
-}
-
-func (rc *refCount) removeSet(ipsetName Name) {
-	for k := range rc.ref {
-		if k.ipsetName == ipsetName {
-			delete(rc.ref, k)
-		}
-	}
-}

npc/iptables/iptables.go

@@ -0,0 +1,7 @@
+package iptables
+
+type Interface interface {
+	Append(table, chain string, rulespec ...string) error
+	Delete(table, chain string, rulespec ...string) error
+	Insert(table, chain string, pos int, rulespec ...string) error
+}

npc/namespace.go

@@ -2,37 +2,50 @@ package npc
 
 import (
 	"encoding/json"
+	"errors"
+	"fmt"
 
-	"github.com/coreos/go-iptables/iptables"
-	"k8s.io/client-go/pkg/api/unversioned"
-	coreapi "k8s.io/client-go/pkg/api/v1"
-	extnapi "k8s.io/client-go/pkg/apis/extensions/v1beta1"
-	"k8s.io/client-go/pkg/types"
-	"k8s.io/client-go/pkg/util/uuid"
+	coreapi "k8s.io/api/core/v1"
+	extnapi "k8s.io/api/extensions/v1beta1"
+	networkingv1 "k8s.io/api/networking/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/types"
+	"k8s.io/apimachinery/pkg/util/uuid"
 
 	"github.com/weaveworks/weave/common"
 	"github.com/weaveworks/weave/npc/ipset"
+	"github.com/weaveworks/weave/npc/iptables"
 )
 
+var errInvalidNetworkPolicyObjType = errors.New("invalid NetworkPolicy object type")
+
 type ns struct {
-	ipt *iptables.IPTables // interface to iptables
+	ipt iptables.Interface // interface to iptables
 	ips ipset.Interface    // interface to ipset
 
-	name      string                               // k8s Namespace name
-	namespace *coreapi.Namespace                   // k8s Namespace object
-	pods      map[types.UID]*coreapi.Pod           // k8s Pod objects by UID
-	policies  map[types.UID]*extnapi.NetworkPolicy // k8s NetworkPolicy objects by UID
+	name      string                     // k8s Namespace name
+	nodeName  string                     // my node name
+	namespace *coreapi.Namespace         // k8s Namespace object
+	pods      map[types.UID]*coreapi.Pod // k8s Pod objects by UID
+	policies  map[types.UID]interface{}  // k8s NetworkPolicy objects by UID
 
 	uid     types.UID     // surrogate UID to own allPods selector
 	allPods *selectorSpec // hash:ip ipset of all pod IPs in this namespace
 
+	// stores IP addrs of pods which are not selected by any dst podSelector of
+	// any netpol; used only in non-legacy mode and is used as a dst in
+	// the WEAVE-NPC-DEFAULT iptables chain.
+	defaultAllowIPSet ipset.Name
+
 	nsSelectors  *selectorSet
 	podSelectors *selectorSet
 	rules        *ruleSet
+
+	legacy bool
 }
 
-func newNS(name string, ipt *iptables.IPTables, ips ipset.Interface, nsSelectors *selectorSet) (*ns, error) {
-	allPods, err := newSelectorSpec(&unversioned.LabelSelector{}, name, ipset.HashIP)
+func newNS(name, nodeName string, legacy bool, ipt iptables.Interface, ips ipset.Interface, nsSelectors *selectorSet) (*ns, error) {
+	allPods, err := newSelectorSpec(&metav1.LabelSelector{}, false, name, ipset.HashIP)
 	if err != nil {
 		return nil, err
 	}
@@ -41,14 +54,24 @@ func newNS(name string, ipt *iptables.IPTables, ips ipset.Interface, nsSelectors
 		ipt:         ipt,
 		ips:         ips,
 		name:        name,
+		nodeName:    nodeName,
 		pods:        make(map[types.UID]*coreapi.Pod),
-		policies:    make(map[types.UID]*extnapi.NetworkPolicy),
+		policies:    make(map[types.UID]interface{}),
 		uid:         uuid.NewUUID(),
 		allPods:     allPods,
 		nsSelectors: nsSelectors,
-		rules:       newRuleSet(ipt)}
+		rules:       newRuleSet(ipt),
+		legacy:      legacy}
 
-	ns.podSelectors = newSelectorSet(ips, ns.onNewPodSelector)
+	ns.podSelectors = newSelectorSet(ips, ns.onNewPodSelector, ns.onNewDstPodSelector, ns.onDestroyDstPodSelector)
+
+	if !legacy {
+		defaultAllowIPSet := ipset.Name(IpsetNamePrefix + shortName("default-allow:"+name))
+		if err := ips.Create(defaultAllowIPSet, ipset.HashIP); err != nil {
+			return nil, err
+		}
+		ns.defaultAllowIPSet = defaultAllowIPSet
+	}
 
 	if err := ns.podSelectors.provision(ns.uid, nil, map[string]*selectorSpec{ns.allPods.key: ns.allPods}); err != nil {
 		return nil, err
@@ -62,25 +85,85 @@ func (ns *ns) empty() bool {
 }
 
 func (ns *ns) destroy() error {
-	if err := ns.podSelectors.deprovision(ns.uid, map[string]*selectorSpec{ns.allPods.key: ns.allPods}, nil); err != nil {
-		return err
-	}
-	return nil
+	return ns.podSelectors.deprovision(ns.uid, map[string]*selectorSpec{ns.allPods.key: ns.allPods}, nil)
 }
 
 func (ns *ns) onNewPodSelector(selector *selector) error {
 	for _, pod := range ns.pods {
 		if hasIP(pod) {
 			if selector.matches(pod.ObjectMeta.Labels) {
-				if err := selector.addEntry(pod.Status.PodIP); err != nil {
+				if err := selector.addEntry(pod.ObjectMeta.UID, pod.Status.PodIP, podComment(pod)); err != nil {
+					return err
+				}
+
+			}
+		}
+	}
+	return nil
+}
+
+func (ns *ns) onNewDstPodSelector(selector *selector) error {
+	if ns.legacy {
+		return nil
+	}
+
+	for _, pod := range ns.pods {
+		if hasIP(pod) {
+			// Remove the pod from default-allow if dst podselector matches the pod
+			if selector.matches(pod.ObjectMeta.Labels) {
+				if err := ns.ips.DelEntry(pod.ObjectMeta.UID, ns.defaultAllowIPSet, pod.Status.PodIP); err != nil {
 					return err
 				}
 			}
 		}
 	}
+
 	return nil
 }
 
+func (ns *ns) onDestroyDstPodSelector(selector *selector) error {
+	if ns.legacy {
+		return nil
+	}
+
+	for _, pod := range ns.pods {
+		if hasIP(pod) {
+			if selector.matches(pod.ObjectMeta.Labels) {
+				if err := ns.addToDefaultAllowIfNoMatching(pod); err != nil {
+					return err
+				}
+			}
+		}
+	}
+
+	return nil
+}
+
+// Add pod IP addr to default-allow ipset if there are no matching dst selectors
+func (ns *ns) addToDefaultAllowIfNoMatching(pod *coreapi.Pod) error {
+	found := false
+	// TODO(mp) optimize (avoid iterating over selectors) by ref counting IP addrs.
+	for _, s := range ns.podSelectors.entries {
+		if ns.podSelectors.dstSelectorExist(s) && s.matches(pod.ObjectMeta.Labels) {
+			found = true
+			break
+		}
+	}
+	if !found {
+		if err := ns.ips.AddEntry(pod.ObjectMeta.UID, ns.defaultAllowIPSet, pod.Status.PodIP, podComment(pod)); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+func (ns *ns) checkLocalPod(obj *coreapi.Pod) bool {
+	if obj.Spec.NodeName != ns.nodeName {
+		return false
+	}
+	return true
+}
+
 func (ns *ns) addPod(obj *coreapi.Pod) error {
 	ns.pods[obj.ObjectMeta.UID] = obj
 
@@ -88,8 +171,22 @@ func (ns *ns) addPod(obj *coreapi.Pod) error {
 		return nil
 	}
 
-	ns.ips.AddEntry(LocalIpset, obj.Status.PodIP)
-	return ns.podSelectors.addToMatching(obj.ObjectMeta.Labels, obj.Status.PodIP)
+	if ns.checkLocalPod(obj) {
+		ns.ips.AddEntry(obj.ObjectMeta.UID, LocalIpset, obj.Status.PodIP, podComment(obj))
+	}
+
+	found, err := ns.podSelectors.addToMatching(obj.ObjectMeta.UID, obj.ObjectMeta.Labels, obj.Status.PodIP, podComment(obj))
+	if err != nil {
+		return err
+	}
+	// If there are no matching dst selectors, add the pod to default-allow
+	if !ns.legacy && !found {
+		if err := ns.ips.AddEntry(obj.ObjectMeta.UID, ns.defaultAllowIPSet, obj.Status.PodIP, podComment(obj)); err != nil {
+			return err
+		}
+	}
+
+	return nil
 }
 
 func (ns *ns) updatePod(oldObj, newObj *coreapi.Pod) error {
@@ -101,13 +198,50 @@ func (ns *ns) updatePod(oldObj, newObj *coreapi.Pod) error {
 	}
 
 	if hasIP(oldObj) && !hasIP(newObj) {
-		ns.ips.DelEntry(LocalIpset, oldObj.Status.PodIP)
-		return ns.podSelectors.delFromMatching(oldObj.ObjectMeta.Labels, oldObj.Status.PodIP)
+		if ns.checkLocalPod(oldObj) {
+			ns.ips.DelEntry(oldObj.ObjectMeta.UID, LocalIpset, oldObj.Status.PodIP)
+		}
+
+		if !ns.legacy {
+			if err := ns.ips.DelEntry(oldObj.ObjectMeta.UID, ns.defaultAllowIPSet, oldObj.Status.PodIP); err != nil {
+				return err
+			}
+		}
+
+		return ns.podSelectors.delFromMatching(oldObj.ObjectMeta.UID, oldObj.ObjectMeta.Labels, oldObj.Status.PodIP)
 	}
 
 	if !hasIP(oldObj) && hasIP(newObj) {
-		ns.ips.AddEntry(LocalIpset, newObj.Status.PodIP)
-		return ns.podSelectors.addToMatching(newObj.ObjectMeta.Labels, newObj.Status.PodIP)
+		if ns.checkLocalPod(newObj) {
+			ns.ips.AddEntry(newObj.ObjectMeta.UID, LocalIpset, newObj.Status.PodIP, podComment(newObj))
+		}
+		found, err := ns.podSelectors.addToMatching(newObj.ObjectMeta.UID, newObj.ObjectMeta.Labels, newObj.Status.PodIP, podComment(newObj))
+		if err != nil {
+			return err
+		}
+
+		if !ns.legacy && !found {
+			if err := ns.ips.AddEntry(newObj.ObjectMeta.UID, ns.defaultAllowIPSet, newObj.Status.PodIP, podComment(newObj)); err != nil {
+				return err
+			}
+		}
+
+		return nil
+	}
+
+	if !ns.legacy &&
+		oldObj.Status.PodIP != newObj.Status.PodIP &&
+		ns.ips.Exist(oldObj.ObjectMeta.UID, ns.defaultAllowIPSet, oldObj.Status.PodIP) {
+		// Instead of iterating over all selectors we check whether old pod IP
+		// has been inserted into default-allow ipset to decide whether the IP
+		// in the ipset has to be updated.
+
+		if err := ns.ips.DelEntry(oldObj.ObjectMeta.UID, ns.defaultAllowIPSet, oldObj.Status.PodIP); err != nil {
+			return err
+		}
+		if err := ns.ips.AddEntry(newObj.ObjectMeta.UID, ns.defaultAllowIPSet, newObj.Status.PodIP, podComment(newObj)); err != nil {
+			return err
+		}
 	}
 
 	if !equals(oldObj.ObjectMeta.Labels, newObj.ObjectMeta.Labels) ||
@@ -120,15 +254,28 @@ func (ns *ns) updatePod(oldObj, newObj *coreapi.Pod) error {
 				continue
 			}
 			if oldMatch {
-				if err := ps.delEntry(oldObj.Status.PodIP); err != nil {
+				if err := ps.delEntry(oldObj.ObjectMeta.UID, oldObj.Status.PodIP); err != nil {
 					return err
 				}
 			}
 			if newMatch {
-				if err := ps.addEntry(newObj.Status.PodIP); err != nil {
+				if err := ps.addEntry(newObj.ObjectMeta.UID, newObj.Status.PodIP, podComment(newObj)); err != nil {
 					return err
 				}
 			}
+
+			if !ns.legacy && ns.podSelectors.dstSelectorExist(ps) {
+				switch {
+				case !oldMatch && newMatch:
+					if err := ns.ips.DelEntry(oldObj.ObjectMeta.UID, ns.defaultAllowIPSet, oldObj.Status.PodIP); err != nil {
+						return err
+					}
+				case oldMatch && !newMatch:
+					if err := ns.addToDefaultAllowIfNoMatching(newObj); err != nil {
+						return err
+					}
+				}
+			}
 		}
 	}
 
@@ -142,130 +289,150 @@ func (ns *ns) deletePod(obj *coreapi.Pod) error {
 		return nil
 	}
 
-	ns.ips.DelEntry(LocalIpset, obj.Status.PodIP)
-	return ns.podSelectors.delFromMatching(obj.ObjectMeta.Labels, obj.Status.PodIP)
+	if ns.checkLocalPod(obj) {
+		ns.ips.DelEntry(obj.ObjectMeta.UID, LocalIpset, obj.Status.PodIP)
+	}
+
+	if !ns.legacy {
+		if err := ns.ips.DelEntry(obj.ObjectMeta.UID, ns.defaultAllowIPSet, obj.Status.PodIP); err != nil {
+			return err
+		}
+	}
+
+	return ns.podSelectors.delFromMatching(obj.ObjectMeta.UID, obj.ObjectMeta.Labels, obj.Status.PodIP)
 }
 
-func (ns *ns) addNetworkPolicy(obj *extnapi.NetworkPolicy) error {
-	ns.policies[obj.ObjectMeta.UID] = obj
-
+func (ns *ns) addNetworkPolicy(obj interface{}) error {
 	// Analyse policy, determine which rules and ipsets are required
-	rules, nsSelectors, podSelectors, err := ns.analysePolicy(obj)
+
+	uid, rules, nsSelectors, podSelectors, err := ns.analyse(obj)
 	if err != nil {
 		return err
 	}
 
 	// Provision required resources in dependency order
-	if err := ns.nsSelectors.provision(obj.ObjectMeta.UID, nil, nsSelectors); err != nil {
+	if err := ns.nsSelectors.provision(uid, nil, nsSelectors); err != nil {
 		return err
 	}
-	if err := ns.podSelectors.provision(obj.ObjectMeta.UID, nil, podSelectors); err != nil {
+	if err := ns.podSelectors.provision(uid, nil, podSelectors); err != nil {
 		return err
 	}
-	if err := ns.rules.provision(obj.ObjectMeta.UID, nil, rules); err != nil {
-		return err
-	}
-
-	return nil
+	return ns.rules.provision(uid, nil, rules)
 }
 
-func (ns *ns) updateNetworkPolicy(oldObj, newObj *extnapi.NetworkPolicy) error {
-	delete(ns.policies, oldObj.ObjectMeta.UID)
-	ns.policies[newObj.ObjectMeta.UID] = newObj
-
+func (ns *ns) updateNetworkPolicy(oldObj, newObj interface{}) error {
 	// Analyse the old and the new policy so we can determine differences
-	oldRules, oldNsSelectors, oldPodSelectors, err := ns.analysePolicy(oldObj)
+	oldUID, oldRules, oldNsSelectors, oldPodSelectors, err := ns.analyse(oldObj)
 	if err != nil {
 		return err
 	}
-	newRules, newNsSelectors, newPodSelectors, err := ns.analysePolicy(newObj)
+	newUID, newRules, newNsSelectors, newPodSelectors, err := ns.analyse(newObj)
 	if err != nil {
 		return err
 	}
 
+	delete(ns.policies, oldUID)
+	ns.policies[newUID] = newObj
+
 	// Deprovision unused and provision newly required resources in dependency order
-	if err := ns.rules.deprovision(oldObj.ObjectMeta.UID, oldRules, newRules); err != nil {
+	if err := ns.rules.deprovision(oldUID, oldRules, newRules); err != nil {
 		return err
 	}
-	if err := ns.nsSelectors.deprovision(oldObj.ObjectMeta.UID, oldNsSelectors, newNsSelectors); err != nil {
+	if err := ns.nsSelectors.deprovision(oldUID, oldNsSelectors, newNsSelectors); err != nil {
 		return err
 	}
-	if err := ns.podSelectors.deprovision(oldObj.ObjectMeta.UID, oldPodSelectors, newPodSelectors); err != nil {
+	if err := ns.podSelectors.deprovision(oldUID, oldPodSelectors, newPodSelectors); err != nil {
 		return err
 	}
-	if err := ns.nsSelectors.provision(oldObj.ObjectMeta.UID, oldNsSelectors, newNsSelectors); err != nil {
+	if err := ns.nsSelectors.provision(oldUID, oldNsSelectors, newNsSelectors); err != nil {
 		return err
 	}
-	if err := ns.podSelectors.provision(oldObj.ObjectMeta.UID, oldPodSelectors, newPodSelectors); err != nil {
+	if err := ns.podSelectors.provision(oldUID, oldPodSelectors, newPodSelectors); err != nil {
 		return err
 	}
-	if err := ns.rules.provision(oldObj.ObjectMeta.UID, oldRules, newRules); err != nil {
-		return err
-	}
-
-	return nil
+	return ns.rules.provision(oldUID, oldRules, newRules)
 }
 
-func (ns *ns) deleteNetworkPolicy(obj *extnapi.NetworkPolicy) error {
-	delete(ns.policies, obj.ObjectMeta.UID)
-
+func (ns *ns) deleteNetworkPolicy(obj interface{}) error {
 	// Analyse network policy to free resources
-	rules, nsSelectors, podSelectors, err := ns.analysePolicy(obj)
+	uid, rules, nsSelectors, podSelectors, err := ns.analyse(obj)
 	if err != nil {
 		return err
 	}
 
+	delete(ns.policies, uid)
+
 	// Deprovision unused resources in dependency order
-	if err := ns.rules.deprovision(obj.ObjectMeta.UID, rules, nil); err != nil {
+	if err := ns.rules.deprovision(uid, rules, nil); err != nil {
 		return err
 	}
-	if err := ns.nsSelectors.deprovision(obj.ObjectMeta.UID, nsSelectors, nil); err != nil {
+	if err := ns.nsSelectors.deprovision(uid, nsSelectors, nil); err != nil {
 		return err
 	}
-	if err := ns.podSelectors.deprovision(obj.ObjectMeta.UID, podSelectors, nil); err != nil {
-		return err
+	return ns.podSelectors.deprovision(uid, podSelectors, nil)
+}
+
+func bypassRule(nsIpsetName ipset.Name, namespace string) []string {
+	return []string{"-m", "set", "--match-set", string(nsIpsetName), "dst", "-j", "ACCEPT", "-m", "comment", "--comment", "DefaultAllow isolation for namespace: " + namespace}
+}
+
+func (ns *ns) ensureBypassRule() error {
+	var ipset ipset.Name
+	if ns.legacy {
+		ipset = ns.allPods.ipsetName
+	} else {
+		ipset = ns.defaultAllowIPSet
 	}
 
-	return nil
+	common.Log.Debugf("ensuring rule for DefaultAllow in namespace: %s, set %s", ns.name, ipset)
+	return ns.ipt.Append(TableFilter, DefaultChain, bypassRule(ipset, ns.name)...)
 }
 
-func bypassRule(nsIpsetName ipset.Name) []string {
-	return []string{"-m", "set", "--match-set", string(nsIpsetName), "dst", "-j", "ACCEPT"}
-}
+func (ns *ns) deleteBypassRule() error {
+	var ipset ipset.Name
+	if ns.legacy {
+		ipset = ns.allPods.ipsetName
+	} else {
+		ipset = ns.defaultAllowIPSet
+	}
 
-func (ns *ns) ensureBypassRule(nsIpsetName ipset.Name) error {
-	return ns.ipt.Append(TableFilter, DefaultChain, bypassRule(ns.allPods.ipsetName)...)
-}
-
-func (ns *ns) deleteBypassRule(nsIpsetName ipset.Name) error {
-	return ns.ipt.Delete(TableFilter, DefaultChain, bypassRule(ns.allPods.ipsetName)...)
+	common.Log.Debugf("removing default rule in namespace: %s, set %s", ns.name, ipset)
+	return ns.ipt.Delete(TableFilter, DefaultChain, bypassRule(ipset, ns.name)...)
 }
 
 func (ns *ns) addNamespace(obj *coreapi.Namespace) error {
 	ns.namespace = obj
 
 	// Insert a rule to bypass policies if namespace is DefaultAllow
-	if !isDefaultDeny(obj) {
-		return ns.ensureBypassRule(ns.allPods.ipsetName)
+	if !ns.isDefaultDeny(obj) {
+		if err := ns.ensureBypassRule(); err != nil {
+			return err
+		}
 	}
 
 	// Add namespace ipset to matching namespace selectors
-	return ns.nsSelectors.addToMatching(obj.ObjectMeta.Labels, string(ns.allPods.ipsetName))
+	_, err := ns.nsSelectors.addToMatching(obj.ObjectMeta.UID, obj.ObjectMeta.Labels, string(ns.allPods.ipsetName), namespaceComment(ns))
+	return err
 }
 
 func (ns *ns) updateNamespace(oldObj, newObj *coreapi.Namespace) error {
 	ns.namespace = newObj
 
 	// Update bypass rule if ingress default has changed
-	oldDefaultDeny := isDefaultDeny(oldObj)
-	newDefaultDeny := isDefaultDeny(newObj)
+	oldDefaultDeny := ns.isDefaultDeny(oldObj)
+	newDefaultDeny := ns.isDefaultDeny(newObj)
 
 	if oldDefaultDeny != newDefaultDeny {
+		common.Log.Infof("namespace DefaultDeny changed from %t to %t", oldDefaultDeny, newDefaultDeny)
 		if oldDefaultDeny {
-			return ns.ensureBypassRule(ns.allPods.ipsetName)
+			if err := ns.ensureBypassRule(); err != nil {
+				return err
+			}
 		}
 		if newDefaultDeny {
-			return ns.deleteBypassRule(ns.allPods.ipsetName)
+			if err := ns.deleteBypassRule(); err != nil {
+				return err
+			}
 		}
 	}
 
@@ -278,12 +445,12 @@ func (ns *ns) updateNamespace(oldObj, newObj *coreapi.Namespace) error {
 				continue
 			}
 			if oldMatch {
-				if err := selector.delEntry(string(ns.allPods.ipsetName)); err != nil {
+				if err := selector.delEntry(ns.namespace.ObjectMeta.UID, string(ns.allPods.ipsetName)); err != nil {
 					return err
 				}
 			}
 			if newMatch {
-				if err := selector.addEntry(string(ns.allPods.ipsetName)); err != nil {
+				if err := selector.addEntry(ns.namespace.ObjectMeta.UID, string(ns.allPods.ipsetName), namespaceComment(ns)); err != nil {
 					return err
 				}
 			}
@@ -297,12 +464,37 @@ func (ns *ns) deleteNamespace(obj *coreapi.Namespace) error {
 	ns.namespace = nil
 
 	// Remove bypass rule
-	if !isDefaultDeny(obj) {
-		return ns.deleteBypassRule(ns.allPods.ipsetName)
+	if !ns.isDefaultDeny(obj) {
+		if err := ns.deleteBypassRule(); err != nil {
+			return err
+		}
 	}
 
 	// Remove namespace ipset from any matching namespace selectors
-	return ns.nsSelectors.delFromMatching(obj.ObjectMeta.Labels, string(ns.allPods.ipsetName))
+	return ns.nsSelectors.delFromMatching(obj.ObjectMeta.UID, obj.ObjectMeta.Labels, string(ns.allPods.ipsetName))
+}
+
+func (ns *ns) isDefaultDeny(namespace *coreapi.Namespace) bool {
+	nnpJSON, found := namespace.ObjectMeta.Annotations["net.beta.kubernetes.io/network-policy"]
+	if !found {
+		return false
+	}
+
+	if !ns.legacy {
+		common.Log.Warn("DefaultDeny annotation is supported only in legacy mode (--use-legacy-netpol)")
+		return false
+	}
+
+	var nnp NamespaceNetworkPolicy
+	if err := json.Unmarshal([]byte(nnpJSON), &nnp); err != nil {
+		common.Log.Warn("Ignoring network policy annotation: unmarshal failed:", err)
+		// If we can't understand the annotation, behave as if it isn't present
+		return false
+	}
+
+	return nnp.Ingress != nil &&
+		nnp.Ingress.Isolation != nil &&
+		*(nnp.Ingress.Isolation) == DefaultDeny
 }
 
 func hasIP(pod *coreapi.Pod) bool {
@@ -323,20 +515,43 @@ func equals(a, b map[string]string) bool {
 	return true
 }
 
-func isDefaultDeny(namespace *coreapi.Namespace) bool {
-	nnpJSON, found := namespace.ObjectMeta.Annotations["net.beta.kubernetes.io/network-policy"]
-	if !found {
-		return false
+func namespaceComment(namespace *ns) string {
+	return "namespace: " + namespace.name
+}
+
+func podComment(pod *coreapi.Pod) string {
+	return fmt.Sprintf("namespace: %s, pod: %s", pod.ObjectMeta.Namespace, pod.ObjectMeta.Name)
+}
+
+func (ns *ns) analyse(obj interface{}) (
+	uid types.UID,
+	rules map[string]*ruleSpec,
+	nsSelectors, podSelectors map[string]*selectorSpec,
+	err error) {
+
+	switch p := obj.(type) {
+	case *extnapi.NetworkPolicy:
+		uid = p.ObjectMeta.UID
+	case *networkingv1.NetworkPolicy:
+		uid = p.ObjectMeta.UID
+	default:
+		err = errInvalidNetworkPolicyObjType
+		return
 	}
+	ns.policies[uid] = obj
 
-	var nnp NamespaceNetworkPolicy
-	if err := json.Unmarshal([]byte(nnpJSON), &nnp); err != nil {
-		common.Log.Warn("Ignoring network policy annotation: unmarshal failed:", err)
-		// If we can't understand the annotation, behave as if it isn't present
-		return false
+	// Analyse policy, determine which rules and ipsets are required
+	if ns.legacy {
+		rules, nsSelectors, podSelectors, err = ns.analysePolicyLegacy(obj.(*extnapi.NetworkPolicy))
+		if err != nil {
+			return
+		}
+	} else {
+		rules, nsSelectors, podSelectors, err = ns.analysePolicy(obj.(*networkingv1.NetworkPolicy))
+		if err != nil {
+			return
+		}
 	}
 
-	return nnp.Ingress != nil &&
-		nnp.Ingress.Isolation != nil &&
-		*(nnp.Ingress.Isolation) == DefaultDeny
+	return
 }

npc/rule.go

@@ -1,12 +1,13 @@
 package npc
 
 import (
+	"fmt"
 	"strings"
 
-	"github.com/coreos/go-iptables/iptables"
-	"k8s.io/client-go/pkg/types"
+	"k8s.io/apimachinery/pkg/types"
 
 	"github.com/weaveworks/weave/common"
+	"github.com/weaveworks/weave/npc/iptables"
 )
 
 type ruleSpec struct {
@@ -19,27 +20,36 @@ func newRuleSpec(proto *string, srcHost *selectorSpec, dstHost *selectorSpec, ds
 	if proto != nil {
 		args = append(args, "-p", *proto)
 	}
+	srcComment := "anywhere"
 	if srcHost != nil {
 		args = append(args, "-m", "set", "--match-set", string(srcHost.ipsetName), "src")
+		if srcHost.nsName != "" {
+			srcComment = fmt.Sprintf("pods: namespace: %s, selector: %s", srcHost.nsName, srcHost.key)
+		} else {
+			srcComment = fmt.Sprintf("namespaces: selector: %s", srcHost.key)
+		}
 	}
+	dstComment := "anywhere"
 	if dstHost != nil {
 		args = append(args, "-m", "set", "--match-set", string(dstHost.ipsetName), "dst")
+		dstComment = fmt.Sprintf("pods: namespace: %s, selector: %s", dstHost.nsName, dstHost.key)
 	}
 	if dstPort != nil {
 		args = append(args, "--dport", *dstPort)
 	}
 	args = append(args, "-j", "ACCEPT")
+	args = append(args, "-m", "comment", "--comment", fmt.Sprintf("%s -> %s", srcComment, dstComment))
 	key := strings.Join(args, " ")
 
 	return &ruleSpec{key, args}
 }
 
 type ruleSet struct {
-	ipt   *iptables.IPTables
+	ipt   iptables.Interface
 	users map[string]map[types.UID]struct{}
 }
 
-func newRuleSet(ipt *iptables.IPTables) *ruleSet {
+func newRuleSet(ipt iptables.Interface) *ruleSet {
 	return &ruleSet{ipt, make(map[string]map[types.UID]struct{})}
 }
 

npc/rule_test.go

@@ -0,0 +1,16 @@
+package npc
+
+import (
+	"github.com/stretchr/testify/require"
+	"testing"
+)
+
+func TestRegressionHandleNil3095(t *testing.T) {
+	// Test for handling nil values
+	// https://github.com/weaveworks/weave/issues/3095
+
+	rs := newRuleSpec(nil, nil, nil, nil)
+
+	require.NotEqual(t, rs.key, "")
+
+}

npc/selector.go

@@ -1,9 +1,9 @@
 package npc
 
 import (
-	"k8s.io/client-go/pkg/api/unversioned"
-	"k8s.io/client-go/pkg/labels"
-	"k8s.io/client-go/pkg/types"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/labels"
+	"k8s.io/apimachinery/pkg/types"
 
 	"github.com/weaveworks/weave/common"
 	"github.com/weaveworks/weave/npc/ipset"
@@ -12,13 +12,15 @@ import (
 type selectorSpec struct {
 	key      string          // string representation (for hash keying/equality comparison)
 	selector labels.Selector // k8s Selector object (for matching)
+	dst      bool            // destination selector
 
 	ipsetType ipset.Type // type of ipset to provision
 	ipsetName ipset.Name // generated ipset name
+	nsName    string     // Namespace name
 }
 
-func newSelectorSpec(json *unversioned.LabelSelector, nsName string, ipsetType ipset.Type) (*selectorSpec, error) {
-	selector, err := unversioned.LabelSelectorAsSelector(json)
+func newSelectorSpec(json *metav1.LabelSelector, dst bool, nsName string, ipsetType ipset.Type) (*selectorSpec, error) {
+	selector, err := metav1.LabelSelectorAsSelector(json)
 	if err != nil {
 		return nil, err
 	}
@@ -26,11 +28,13 @@ func newSelectorSpec(json *unversioned.LabelSelector, nsName string, ipsetType i
 	return &selectorSpec{
 		key:      key,
 		selector: selector,
+		dst:      dst,
 		// We prefix the selector string with the namespace name when generating
 		// the shortname because you can specify the same selector in multiple
 		// namespaces - we need those to map to distinct ipsets
 		ipsetName: ipset.Name(IpsetNamePrefix + shortName(nsName+":"+key)),
-		ipsetType: ipsetType}, nil
+		ipsetType: ipsetType,
+		nsName:    nsName}, nil
 }
 
 type selector struct {
@@ -42,12 +46,12 @@ func (s *selector) matches(labelMap map[string]string) bool {
 	return s.spec.selector.Matches(labels.Set(labelMap))
 }
 
-func (s *selector) addEntry(entry string) error {
-	return s.ips.AddEntry(s.spec.ipsetName, entry)
+func (s *selector) addEntry(user types.UID, entry string, comment string) error {
+	return s.ips.AddEntry(user, s.spec.ipsetName, entry, comment)
 }
 
-func (s *selector) delEntry(entry string) error {
-	return s.ips.DelEntry(s.spec.ipsetName, entry)
+func (s *selector) delEntry(user types.UID, entry string) error {
+	return s.ips.DelEntry(user, s.spec.ipsetName, entry)
 }
 
 type selectorFn func(selector *selector) error
@@ -55,22 +59,51 @@ type selectorFn func(selector *selector) error
 type selectorSet struct {
 	ips           ipset.Interface
 	onNewSelector selectorFn
-	users         map[string]map[types.UID]struct{} // list of users per selector
-	entries       map[string]*selector
+
+	// invoked after dst selector has been provisioned for the first time
+	onNewDstSelector selectorFn
+	// invoked after the last instance of dst selector has been deprovisioned
+	onDestroyDstSelector selectorFn
+
+	users   map[string]map[types.UID]struct{} // list of users per selector
+	entries map[string]*selector
+
+	// We need to keep track of dst selector instances to be able to invoke
+	// onNewDstSelector and onDestroyDstSelector callbacks at the right time;
+	// selectorSpec.Key -> count
+	dstSelectorsCount map[string]int
 }
 
-func newSelectorSet(ips ipset.Interface, onNewSelector selectorFn) *selectorSet {
+func newSelectorSet(ips ipset.Interface, onNewSelector, onNewDstSelector selectorFn, onDestroyDstSelector selectorFn) *selectorSet {
 	return &selectorSet{
-		ips:           ips,
-		onNewSelector: onNewSelector,
-		users:         make(map[string]map[types.UID]struct{}),
-		entries:       make(map[string]*selector)}
+		ips:                  ips,
+		onNewSelector:        onNewSelector,
+		onNewDstSelector:     onNewDstSelector,
+		onDestroyDstSelector: onDestroyDstSelector,
+		users:                make(map[string]map[types.UID]struct{}),
+		entries:              make(map[string]*selector),
+		dstSelectorsCount:    make(map[string]int)}
 }
 
-func (ss *selectorSet) addToMatching(labelMap map[string]string, entry string) error {
+func (ss *selectorSet) addToMatching(user types.UID, labelMap map[string]string, entry string, comment string) (bool, error) {
+	found := false
 	for _, s := range ss.entries {
 		if s.matches(labelMap) {
-			if err := s.addEntry(entry); err != nil {
+			if ss.dstSelectorExist(s) {
+				found = true
+			}
+			if err := s.addEntry(user, entry, comment); err != nil {
+				return found, err
+			}
+		}
+	}
+	return found, nil
+}
+
+func (ss *selectorSet) delFromMatching(user types.UID, labelMap map[string]string, entry string) error {
+	for _, s := range ss.entries {
+		if s.matches(labelMap) {
+			if err := s.delEntry(user, entry); err != nil {
 				return err
 			}
 		}
@@ -78,15 +111,8 @@ func (ss *selectorSet) addToMatching(labelMap map[string]string, entry string) e
 	return nil
 }
 
-func (ss *selectorSet) delFromMatching(labelMap map[string]string, entry string) error {
-	for _, s := range ss.entries {
-		if s.matches(labelMap) {
-			if err := s.delEntry(entry); err != nil {
-				return err
-			}
-		}
-	}
-	return nil
+func (ss *selectorSet) dstSelectorExist(s *selector) bool {
+	return ss.dstSelectorsCount[s.spec.key] > 0
 }
 
 func (ss *selectorSet) deprovision(user types.UID, current, desired map[string]*selectorSpec) error {
@@ -98,9 +124,19 @@ func (ss *selectorSet) deprovision(user types.UID, current, desired map[string]*
 				if err := ss.ips.Destroy(spec.ipsetName); err != nil {
 					return err
 				}
+
 				delete(ss.entries, key)
 				delete(ss.users, key)
 			}
+
+			if spec.dst {
+				ss.dstSelectorsCount[key]--
+				if ss.dstSelectorsCount[key] == 0 {
+					if err := ss.onDestroyDstSelector(&selector{ss.ips, spec}); err != nil {
+						return err
+					}
+				}
+			}
 		}
 	}
 	return nil
@@ -109,12 +145,13 @@ func (ss *selectorSet) deprovision(user types.UID, current, desired map[string]*
 func (ss *selectorSet) provision(user types.UID, current, desired map[string]*selectorSpec) error {
 	for key, spec := range desired {
 		if _, found := current[key]; !found {
+			selector := &selector{ss.ips, spec}
+
 			if _, found := ss.users[key]; !found {
 				common.Log.Infof("creating ipset: %#v", spec)
 				if err := ss.ips.Create(spec.ipsetName, spec.ipsetType); err != nil {
 					return err
 				}
-				selector := &selector{ss.ips, spec}
 				if err := ss.onNewSelector(selector); err != nil {
 					return err
 				}
@@ -122,6 +159,15 @@ func (ss *selectorSet) provision(user types.UID, current, desired map[string]*se
 				ss.entries[key] = selector
 			}
 			ss.users[key][user] = struct{}{}
+
+			if spec.dst {
+				ss.dstSelectorsCount[key]++
+				if ss.dstSelectorsCount[key] == 1 {
+					if err := ss.onNewDstSelector(selector); err != nil {
+						return err
+					}
+				}
+			}
 		}
 	}
 	return nil

plugin/ipam/cni.go

@@ -38,14 +38,14 @@ func (i *Ipam) Allocate(args *skel.CmdArgs) (types.Result, error) {
 	var ipnet *net.IPNet
 
 	if conf.Subnet == "" {
-		ipnet, err = i.weave.AllocateIP(containerID)
+		ipnet, err = i.weave.AllocateIP(containerID, false)
 	} else {
 		var subnet *net.IPNet
 		subnet, err = types.ParseCIDR(conf.Subnet)
 		if err != nil {
 			return nil, fmt.Errorf("subnet given in config, but not parseable: %s", err)
 		}
-		ipnet, err = i.weave.AllocateIPInSubnet(containerID, subnet)
+		ipnet, err = i.weave.AllocateIPInSubnet(containerID, subnet, false)
 	}
 
 	if err != nil {

plugin/ipam/driver.go

@@ -19,6 +19,11 @@ func NewIpam(weave *api.Client) *Ipam {
 	return &Ipam{weave: weave}
 }
 
+func (i *Ipam) GetCapabilities() (RequiresMACAddress bool, RequiresRequestReplay bool, err error) {
+	i.logReq("GetCapabilities")
+	return false, false, nil
+}
+
 func (i *Ipam) GetDefaultAddressSpaces() (string, string, error) {
 	i.logReq("GetDefaultAddressSpaces")
 	return "weavelocal", "weaveglobal", nil
@@ -73,7 +78,7 @@ func (i *Ipam) RequestAddress(poolID string, address net.IP, options map[string]
 	i.logReq("RequestAddress", poolID, address, options)
 	defer func() { i.logRes("RequestAddress", err, ip) }()
 	if poolID == "weavepool" { // old-style
-		ip, err = i.weave.AllocateIP(api.NoContainerID)
+		ip, err = i.weave.AllocateIP(api.NoContainerID, false)
 		return
 	}
 	subnet, iprange, err := splitPoolID(poolID)
@@ -82,12 +87,12 @@ func (i *Ipam) RequestAddress(poolID string, address net.IP, options map[string]
 	}
 	if address != nil { // try to claim specific address requested
 		ip = &net.IPNet{IP: address, Mask: subnet.Mask}
-		if err = i.weave.ClaimIP(api.NoContainerID, ip); err != nil {
+		if err = i.weave.ClaimIP(api.NoContainerID, ip, false); err != nil {
 			return
 		}
 	} else {
 		// We are lying slightly to IPAM here: the range is not a subnet
-		if ip, err = i.weave.AllocateIPInSubnet(api.NoContainerID, iprange); err != nil {
+		if ip, err = i.weave.AllocateIPInSubnet(api.NoContainerID, iprange, false); err != nil {
 			return
 		}
 		ip.Mask = subnet.Mask // fix up the subnet we lied about

plugin/net/cni.go

@@ -6,7 +6,6 @@ import (
 	"fmt"
 	"net"
 	"os"
-	"syscall"
 
 	"github.com/containernetworking/cni/pkg/ipam"
 	"github.com/containernetworking/cni/pkg/skel"
@@ -100,12 +99,11 @@ func (c *CNIPlugin) CmdAdd(args *skel.CmdArgs) error {
 				return fmt.Errorf("unable to allocate IP address for bridge: %s", err)
 			}
 			bridgeCIDR := bridgeIPResult.IPs[0].Address
-			if err := assignBridgeIP(conf.BrName, bridgeCIDR); err != nil {
-				return fmt.Errorf("unable to assign IP address to bridge: %s", err)
-			}
-			if err := weavenet.ExposeNAT(bridgeCIDR); err != nil {
-				return fmt.Errorf("unable to create NAT rules: %s", err)
+
+			if err := c.weave.Expose(&bridgeCIDR); err != nil {
+				return fmt.Errorf("unable to expose bridge %q: %s", conf.BrName, err)
 			}
+
 			bridgeIP = bridgeCIDR.IP
 		} else if err != nil {
 			return err
@@ -167,22 +165,6 @@ func setupRoutes(link netlink.Link, name string, ipnet net.IPNet, gw net.IP, rou
 	return nil
 }
 
-func assignBridgeIP(bridgeName string, ipnet net.IPNet) error {
-	link, err := netlink.LinkByName(bridgeName)
-	if err != nil {
-		return err
-	}
-	if err := netlink.AddrAdd(link, &netlink.Addr{IPNet: &ipnet}); err != nil {
-		// Treat as non-error if this address is already there
-		// - maybe another copy of this program just added it
-		if err == syscall.Errno(syscall.EEXIST) {
-			return nil
-		}
-		return fmt.Errorf("failed to add IP address to %q: %v", bridgeName, err)
-	}
-	return nil
-}
-
 // As of CNI 0.5 spec:
 //   "Plugins should generally complete a DEL action without error even if some resources are missing"
 // this method should therefore return nil in most, if not all, cases.
@@ -207,7 +189,9 @@ func (c *CNIPlugin) CmdDel(args *skel.CmdArgs) error {
 	} else {
 		err = ipam.ExecDel(conf.IPAM.Type, args.StdinData)
 	}
-	logOnStderr(fmt.Errorf("unable to release IP address: %s", err))
+	if err != nil {
+		logOnStderr(fmt.Errorf("unable to release IP address: %s", err))
+	}
 	return nil
 }
 

plugin/net/driver.go

@@ -28,25 +28,40 @@ type network struct {
 }
 
 type driver struct {
-	name   string
-	scope  string
-	docker *docker.Client
 	sync.RWMutex
-	networks map[string]network
+	name  string
+	scope string
+	// Docker API is not available for plugin-v2
+	docker     *docker.Client
+	dns        bool
+	isPluginV2 bool
+	// Enable multicast regardless whether multicast opt is passed to Docker;
+	// used only by plugin-v2
+	forceMulticast bool
+	networks       map[string]network
 }
 
-func New(client *docker.Client, weave *weaveapi.Client, name, scope string) (skel.Driver, error) {
+func New(client *docker.Client, weave *weaveapi.Client, name, scope string, dns, isPluginV2, forceMulticast bool) (skel.Driver, error) {
 	driver := &driver{
-		name:     name,
-		scope:    scope,
-		docker:   client,
-		networks: make(map[string]network),
+		name:       name,
+		scope:      scope,
+		docker:     client,
+		dns:        dns,
+		isPluginV2: isPluginV2,
+		// make sure that it's used only by plugin-v2
+		forceMulticast: isPluginV2 && forceMulticast,
+		networks:       make(map[string]network),
 	}
 
-	_, err := NewWatcher(client, weave, driver)
-	if err != nil {
-		return nil, err
+	// Do not start watcher in the case of plugin v2, which prevents us from
+	// configuring arp settings of containers.
+	if client != nil {
+		_, err := NewWatcher(client, weave, driver)
+		if err != nil {
+			return nil, err
+		}
 	}
+
 	return driver, nil
 }
 
@@ -61,12 +76,27 @@ func (driver *driver) GetCapabilities() (*api.GetCapabilityResponse, error) {
 	return caps, nil
 }
 
+// In Swarm mode, CreateNetwork is called on each Swarm node when a new service
+// is created.
 func (driver *driver) CreateNetwork(create *api.CreateNetworkRequest) error {
 	driver.logReq("CreateNetwork", create, create.NetworkID)
 	_, err := driver.setupNetworkInfo(create.NetworkID, true, stringOptions(create))
 	return err
 }
 
+// NetworkAllocate is called on a Swarm node (master) which creates the network.
+// The returned options are passed to CreateNetwork.
+func (driver *driver) NetworkAllocate(alloc *api.AllocateNetworkRequest) (*api.AllocateNetworkResponse, error) {
+	driver.logReq("NetworkAllocate", alloc, alloc.NetworkID)
+	return &api.AllocateNetworkResponse{Options: alloc.Options}, nil
+}
+
+// NetworkFree is called on a Swarm master node which created the network.
+func (driver *driver) NetworkFree(free *api.FreeNetworkRequest) (*api.FreeNetworkResponse, error) {
+	driver.logReq("NetworkFree", free, free.NetworkID)
+	return nil, nil
+}
+
 // Deal with excessively-generic way the options get decoded from JSON
 func stringOptions(create *api.CreateNetworkRequest) map[string]string {
 	if create.Options != nil {
@@ -85,6 +115,7 @@ func stringOptions(create *api.CreateNetworkRequest) map[string]string {
 	return nil
 }
 
+// In Swarm mode, DeleteNetwork is called after a service has been removed.
 func (driver *driver) DeleteNetwork(delreq *api.DeleteNetworkRequest) error {
 	driver.logReq("DeleteNetwork", delreq, delreq.NetworkID)
 	driver.Lock()
@@ -163,12 +194,29 @@ func (driver *driver) JoinEndpoint(j *api.JoinRequest) (*api.JoinResponse, error
 }
 
 func (driver *driver) findNetworkInfo(id string) (network, error) {
+	var network network
+
+	// plugin-v2 does not have access to docker.sock, so we cannot call Docker
+	// API for the network info.
+	if driver.isPluginV2 {
+		// safe to set, as isOurs used only by the watcher which is disabled for plugin-v2
+		network.isOurs = true
+		network.hasMulticastRoute = driver.forceMulticast
+
+		return network, nil
+	}
+
 	driver.Lock()
 	network, found := driver.networks[id]
 	driver.Unlock()
 	if found {
 		return network, nil
 	}
+
+	if driver.docker == nil {
+		return network, fmt.Errorf("Docker client disabled; unable to get network info")
+	}
+
 	info, err := driver.docker.NetworkInfo(id)
 	if err != nil {
 		return network, err

plugin/net/watcher.go

@@ -8,10 +8,6 @@ import (
 	weavenet "github.com/weaveworks/weave/net"
 )
 
-const (
-	WeaveDomain = "weave.local"
-)
-
 type watcher struct {
 	client *docker.Client
 	weave  *weaveapi.Client
@@ -41,11 +37,19 @@ func (w *watcher) ContainerStarted(id string) {
 			continue
 		}
 		if network.isOurs {
-			fqdn := fmt.Sprintf("%s.%s", info.Config.Hostname, info.Config.Domainname)
-			if err := w.weave.RegisterWithDNS(id, fqdn, net.IPAddress); err != nil {
-				w.driver.warn("ContainerStarted", "unable to register %s with weaveDNS: %s", id, err)
+			if w.driver.dns {
+				fqdn := fmt.Sprintf("%s.%s", info.Config.Hostname, info.Config.Domainname)
+				if err := w.weave.RegisterWithDNS(id, fqdn, net.IPAddress); err != nil {
+					w.driver.warn("ContainerStarted", "unable to register %s with weaveDNS: %s", id, err)
+				}
 			}
-			if _, err := weavenet.WithNetNSByPid(info.State.Pid, "configure-arp", weavenet.VethName); err != nil {
+			rootDir := "/"
+			if w.driver.isPluginV2 {
+				// We bind mount host's /proc to /host/proc for plugin-v2
+				rootDir = "/host"
+			}
+			netNSPath := weavenet.NSPathByPidWithRoot(rootDir, info.State.Pid)
+			if _, err := weavenet.WithNetNS(netNSPath, "configure-arp", weavenet.VethName, rootDir); err != nil {
 				w.driver.warn("ContainerStarted", "unable to configure interfaces: %s", err)
 			}
 		}

plugin/plugin.go

@@ -0,0 +1,124 @@
+package plugin
+
+import (
+	"net"
+	"os"
+	"path"
+	"strings"
+
+	"github.com/docker/libnetwork/ipamapi"
+	weaveapi "github.com/weaveworks/weave/api"
+	"github.com/weaveworks/weave/common"
+	"github.com/weaveworks/weave/common/docker"
+	weavenet "github.com/weaveworks/weave/net"
+	ipamplugin "github.com/weaveworks/weave/plugin/ipam"
+	netplugin "github.com/weaveworks/weave/plugin/net"
+	"github.com/weaveworks/weave/plugin/skel"
+)
+
+const (
+	pluginV2Name    = "net-plugin"
+	defaultNetwork  = "weave"
+	MulticastOption = netplugin.MulticastOption
+)
+
+var Log = common.Log
+
+type Config struct {
+	Socket            string
+	MeshSocket        string
+	Enable            bool
+	EnableV2          bool
+	EnableV2Multicast bool
+	DNS               bool
+	DefaultSubnet     string
+}
+
+type Plugin struct {
+	Config
+}
+
+func NewPlugin(config Config) *Plugin {
+	if !config.Enable && !config.EnableV2 {
+		return nil
+	}
+	plugin := &Plugin{Config: config}
+	return plugin
+}
+
+func (plugin *Plugin) Start(weaveAPIAddr string, dockerClient *docker.Client, ready func()) {
+	weave := weaveapi.NewClient(weaveAPIAddr, Log)
+
+	Log.Info("Waiting for Weave API Server...")
+	weave.WaitAPIServer(30)
+	Log.Info("Finished waiting for Weave API Server")
+
+	if err := plugin.run(dockerClient, weave, ready); err != nil {
+		Log.Fatal(err)
+	}
+}
+
+func (plugin *Plugin) run(dockerClient *docker.Client, weave *weaveapi.Client, ready func()) error {
+	endChan := make(chan error, 1)
+
+	if plugin.Socket != "" {
+		globalListener, err := listenAndServe(dockerClient, weave, plugin.Socket, endChan, "global", false, plugin.DNS, plugin.EnableV2, plugin.EnableV2Multicast)
+		if err != nil {
+			return err
+		}
+		defer os.Remove(plugin.Socket)
+		defer globalListener.Close()
+	}
+	if plugin.MeshSocket != "" {
+		meshListener, err := listenAndServe(dockerClient, weave, plugin.MeshSocket, endChan, "local", true, plugin.DNS, plugin.EnableV2, plugin.EnableV2Multicast)
+		if err != nil {
+			return err
+		}
+		defer os.Remove(plugin.MeshSocket)
+		defer meshListener.Close()
+		if !plugin.EnableV2 {
+			Log.Printf("Creating default %q network", defaultNetwork)
+			options := map[string]interface{}{MulticastOption: "true"}
+			dockerClient.EnsureNetwork(defaultNetwork, pluginNameFromAddress(plugin.MeshSocket), plugin.DefaultSubnet, options)
+		}
+	}
+	ready()
+
+	return <-endChan
+}
+
+func listenAndServe(dockerClient *docker.Client, weave *weaveapi.Client, address string, endChan chan<- error, scope string, withIpam, dns bool, isPluginV2, forceMulticast bool) (net.Listener, error) {
+	var name string
+	if isPluginV2 {
+		name = pluginV2Name
+	} else {
+		name = pluginNameFromAddress(address)
+	}
+
+	d, err := netplugin.New(dockerClient, weave, name, scope, dns, isPluginV2, forceMulticast)
+	if err != nil {
+		return nil, err
+	}
+
+	var i ipamapi.Ipam
+	if withIpam {
+		i = ipamplugin.NewIpam(weave)
+	}
+
+	listener, err := weavenet.ListenUnixSocket(address)
+	if err != nil {
+		return nil, err
+	}
+	Log.Printf("Listening on %s for %s scope", address, scope)
+
+	go func() {
+		endChan <- skel.Listen(listener, d, i)
+	}()
+
+	return listener, nil
+}
+
+// Take a socket address like "/run/docker/plugins/weavemesh.sock" and extract the plugin name "weavemesh"
+func pluginNameFromAddress(address string) string {
+	return strings.TrimSuffix(path.Base(address), ".sock")
+}

plugin/skel/listener.go

@@ -26,11 +26,17 @@ type Driver interface {
 	DeleteEndpoint(delete *api.DeleteEndpointRequest) error
 	EndpointInfo(req *api.EndpointInfoRequest) (*api.EndpointInfoResponse, error)
 	JoinEndpoint(j *api.JoinRequest) (response *api.JoinResponse, error error)
+	NetworkAllocate(*api.AllocateNetworkRequest) (*api.AllocateNetworkResponse, error)
+	NetworkFree(*api.FreeNetworkRequest) (*api.FreeNetworkResponse, error)
 	LeaveEndpoint(leave *api.LeaveRequest) error
 	DiscoverNew(discover *api.DiscoveryNotification) error
 	DiscoverDelete(delete *api.DiscoveryNotification) error
 }
 
+type IpamCapability interface {
+	GetCapabilities() (RequiresMACAddress bool, RequiresRequestReplay bool, err error)
+}
+
 type listener struct {
 	d Driver
 	i ipamapi.Ipam
@@ -58,9 +64,12 @@ func Listen(socket net.Listener, driver Driver, ipamDriver ipamapi.Ipam) error {
 		handleMethod(networkReceiver, "EndpointOperInfo", listener.infoEndpoint)
 		handleMethod(networkReceiver, "Join", listener.joinEndpoint)
 		handleMethod(networkReceiver, "Leave", listener.leaveEndpoint)
+		handleMethod(networkReceiver, "AllocateNetwork", listener.networkAllocate)
+		handleMethod(networkReceiver, "FreeNetwork", listener.networkFree)
 	}
 
 	if ipamDriver != nil {
+		handleMethod(ipamReceiver, "GetCapabilities", listener.getIpamCapabilities)
 		handleMethod(ipamReceiver, "GetDefaultAddressSpaces", listener.getDefaultAddressSpaces)
 		handleMethod(ipamReceiver, "RequestPool", listener.requestPool)
 		handleMethod(ipamReceiver, "ReleasePool", listener.releasePool)
@@ -137,7 +146,7 @@ func (listener *listener) createEndpoint(w http.ResponseWriter, r *http.Request)
 func (listener *listener) deleteEndpoint(w http.ResponseWriter, r *http.Request) {
 	var delete api.DeleteEndpointRequest
 	if err := json.NewDecoder(r.Body).Decode(&delete); err != nil {
-		sendError(w, "Could not decode JSON encode payload", http.StatusBadRequest)
+		sendError(w, "Could not decode JSON encoded payload", http.StatusBadRequest)
 		return
 	}
 	emptyOrErrorResponse(w, listener.d.DeleteEndpoint(&delete))
@@ -146,7 +155,7 @@ func (listener *listener) deleteEndpoint(w http.ResponseWriter, r *http.Request)
 func (listener *listener) infoEndpoint(w http.ResponseWriter, r *http.Request) {
 	var req api.EndpointInfoRequest
 	if err := json.NewDecoder(r.Body).Decode(&req); err != nil {
-		sendError(w, "Could not decode JSON encode payload", http.StatusBadRequest)
+		sendError(w, "Could not decode JSON encoded payload", http.StatusBadRequest)
 		return
 	}
 	info, err := listener.d.EndpointInfo(&req)
@@ -156,17 +165,37 @@ func (listener *listener) infoEndpoint(w http.ResponseWriter, r *http.Request) {
 func (listener *listener) joinEndpoint(w http.ResponseWriter, r *http.Request) {
 	var join api.JoinRequest
 	if err := json.NewDecoder(r.Body).Decode(&join); err != nil {
-		sendError(w, "Could not decode JSON encode payload", http.StatusBadRequest)
+		sendError(w, "Could not decode JSON encoded payload", http.StatusBadRequest)
 		return
 	}
 	res, err := listener.d.JoinEndpoint(&join)
 	objectOrErrorResponse(w, res, err)
 }
 
+func (listener *listener) networkAllocate(w http.ResponseWriter, r *http.Request) {
+	var alloc api.AllocateNetworkRequest
+	if err := json.NewDecoder(r.Body).Decode(&alloc); err != nil {
+		sendError(w, "Could not decode JSON encode payload", http.StatusBadRequest)
+		return
+	}
+	res, err := listener.d.NetworkAllocate(&alloc)
+	objectOrErrorResponse(w, res, err)
+}
+
+func (listener *listener) networkFree(w http.ResponseWriter, r *http.Request) {
+	var free api.FreeNetworkRequest
+	if err := json.NewDecoder(r.Body).Decode(&free); err != nil {
+		sendError(w, "Could not decode JSON encode payload", http.StatusBadRequest)
+		return
+	}
+	res, err := listener.d.NetworkFree(&free)
+	objectOrErrorResponse(w, res, err)
+}
+
 func (listener *listener) leaveEndpoint(w http.ResponseWriter, r *http.Request) {
 	var l api.LeaveRequest
 	if err := json.NewDecoder(r.Body).Decode(&l); err != nil {
-		sendError(w, "Could not decode JSON encode payload", http.StatusBadRequest)
+		sendError(w, "Could not decode JSON encoded payload", http.StatusBadRequest)
 		return
 	}
 	emptyOrErrorResponse(w, listener.d.LeaveEndpoint(&l))
@@ -175,7 +204,7 @@ func (listener *listener) leaveEndpoint(w http.ResponseWriter, r *http.Request)
 func (listener *listener) discoverNew(w http.ResponseWriter, r *http.Request) {
 	var disco api.DiscoveryNotification
 	if err := json.NewDecoder(r.Body).Decode(&disco); err != nil {
-		sendError(w, "Could not decode JSON encode payload", http.StatusBadRequest)
+		sendError(w, "Could not decode JSON encoded payload", http.StatusBadRequest)
 		return
 	}
 	emptyOrErrorResponse(w, listener.d.DiscoverNew(&disco))
@@ -184,7 +213,7 @@ func (listener *listener) discoverNew(w http.ResponseWriter, r *http.Request) {
 func (listener *listener) discoverDelete(w http.ResponseWriter, r *http.Request) {
 	var disco api.DiscoveryNotification
 	if err := json.NewDecoder(r.Body).Decode(&disco); err != nil {
-		sendError(w, "Could not decode JSON encode payload", http.StatusBadRequest)
+		sendError(w, "Could not decode JSON encoded payload", http.StatusBadRequest)
 		return
 	}
 	emptyOrErrorResponse(w, listener.d.DiscoverDelete(&disco))
@@ -192,6 +221,20 @@ func (listener *listener) discoverDelete(w http.ResponseWriter, r *http.Request)
 
 // ===
 
+func (listener *listener) getIpamCapabilities(w http.ResponseWriter, r *http.Request) {
+	c, ok := listener.i.(IpamCapability)
+	if !ok {
+		http.NotFound(w, r)
+		return
+	}
+	requiresMACAddress, requiresRequestReplay, err := c.GetCapabilities()
+	response := &ipamapi.Capability{
+		RequiresMACAddress:    requiresMACAddress,
+		RequiresRequestReplay: requiresRequestReplay,
+	}
+	objectOrErrorResponse(w, response, err)
+}
+
 func (listener *listener) getDefaultAddressSpaces(w http.ResponseWriter, r *http.Request) {
 	local, global, err := listener.i.GetDefaultAddressSpaces()
 	response := &iapi.GetAddressSpacesResponse{

plugin/status.go

@@ -0,0 +1,19 @@
+package plugin
+
+type Status struct {
+	Config         Config
+	DriverName     string
+	MeshDriverName string `json:"MeshDriverName,omitempty"`
+}
+
+func (plugin *Plugin) NewStatus() *Status {
+	if plugin == nil {
+		return nil
+	}
+	status := &Status{
+		DriverName:     pluginNameFromAddress(plugin.Socket),
+		MeshDriverName: pluginNameFromAddress(plugin.MeshSocket),
+		Config:         plugin.Config,
+	}
+	return status
+}

prog/kube-peers/annotations.go

@@ -0,0 +1,210 @@
+/*
+In order to keep track of active weave peers, we use annotations on the Kubernetes cluster.
+
+Kubernetes uses etcd to distribute and synchronise these annotations so we don't have to.
+
+This module deals with operations on the peerlist backed by Kubernetes' annotation mechanism.
+*/
+package main
+
+import (
+	"encoding/json"
+	"log"
+	"time"
+
+	"github.com/pkg/errors"
+
+	v1 "k8s.io/api/core/v1"
+	kubeErrors "k8s.io/apimachinery/pkg/api/errors"
+	api "k8s.io/apimachinery/pkg/apis/meta/v1"
+	wait "k8s.io/apimachinery/pkg/util/wait"
+	kubernetes "k8s.io/client-go/kubernetes"
+	corev1client "k8s.io/client-go/kubernetes/typed/core/v1"
+)
+
+type configMapAnnotations struct {
+	ConfigMapName string
+	Namespace     string
+	Client        corev1client.ConfigMapsGetter
+	cm            *v1.ConfigMap
+}
+
+func newConfigMapAnnotations(ns string, configMapName string, clientset *kubernetes.Clientset) *configMapAnnotations {
+	return &configMapAnnotations{
+		Namespace:     ns,
+		ConfigMapName: configMapName,
+		Client:        clientset.CoreV1(),
+	}
+}
+
+type peerList struct {
+	Peers []peerInfo
+}
+
+type peerInfo struct {
+	PeerName string // Weave internal unique ID
+	NodeName string // Kubernetes node name
+}
+
+func (pl peerList) contains(peerName string) bool {
+	for _, peer := range pl.Peers {
+		if peer.PeerName == peerName {
+			return true
+		}
+	}
+	return false
+}
+
+func (pl *peerList) add(peerName string, name string) {
+	pl.Peers = append(pl.Peers, peerInfo{PeerName: peerName, NodeName: name})
+}
+
+func (pl *peerList) remove(peerNameToRemove string) {
+	for i := 0; i < len(pl.Peers); {
+		if pl.Peers[i].PeerName == peerNameToRemove {
+			pl.Peers = append(pl.Peers[:i], pl.Peers[i+1:]...)
+		} else {
+			i++
+		}
+	}
+}
+
+const (
+	retryPeriod  = time.Second * 2
+	jitterFactor = 1.0
+
+	// Prefix all our annotation keys with this string so they don't clash with anyone else's
+	KubePeersPrefix = "kube-peers.weave.works/"
+	// KubePeersAnnotationKey is the default annotation key
+	KubePeersAnnotationKey = KubePeersPrefix + "peers"
+)
+
+func (cml *configMapAnnotations) Init() error {
+	for {
+		// Since it's potentially racy to GET, then CREATE if not found, we wrap in a check loop
+		// so that if the configmap is created after our GET but before or CREATE, we'll gracefully
+		// re-try to get the configmap.
+		var err error
+		cml.cm, err = cml.Client.ConfigMaps(cml.Namespace).Get(cml.ConfigMapName, api.GetOptions{})
+		if err != nil {
+			if !kubeErrors.IsNotFound(err) {
+				return errors.Wrapf(err, "Unable to fetch ConfigMap %s/%s", cml.Namespace, cml.ConfigMapName)
+			}
+			cml.cm, err = cml.Client.ConfigMaps(cml.Namespace).Create(&v1.ConfigMap{
+				ObjectMeta: api.ObjectMeta{
+					Name:      cml.ConfigMapName,
+					Namespace: cml.Namespace,
+				},
+			})
+			if err != nil {
+				if kubeErrors.IsAlreadyExists(err) {
+					continue
+				}
+				return errors.Wrapf(err, "Unable to create ConfigMap %s/%s", cml.Namespace, cml.ConfigMapName)
+			}
+		}
+		break
+	}
+	if cml.cm.Annotations == nil {
+		cml.cm.Annotations = make(map[string]string)
+	}
+	return nil
+}
+
+func (cml *configMapAnnotations) GetPeerList() (*peerList, error) {
+	var record peerList
+	if cml.cm == nil {
+		return nil, errors.New("endpoint not initialized, call Init first")
+	}
+	if recordBytes, found := cml.cm.Annotations[KubePeersAnnotationKey]; found {
+		if err := json.Unmarshal([]byte(recordBytes), &record); err != nil {
+			return nil, err
+		}
+	}
+	return &record, nil
+}
+
+func (cml *configMapAnnotations) UpdatePeerList(list peerList) error {
+	recordBytes, err := json.Marshal(list)
+	if err != nil {
+		return err
+	}
+	return cml.UpdateAnnotation(KubePeersAnnotationKey, string(recordBytes))
+}
+
+// Clean up a string so it meets the Kubernetes requiremements for Annotation keys:
+// name part must consist of alphanumeric characters, '-', '_' or '.', and must
+// start and end with an alphanumeric character (e.g. 'MyName', or 'my.name', or '123-abc')
+func cleanKey(key string) string {
+	buf := []byte(key)
+	for i, c := range buf {
+		if (c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z') || (c >= '0' && c <= '9') || c == '-' || c == '_' || c == '.' || c == '/' {
+			continue
+		}
+		buf[i] = '_'
+	}
+	return string(buf)
+}
+
+func (cml *configMapAnnotations) GetAnnotation(key string) (string, bool) {
+	value, ok := cml.cm.Annotations[cleanKey(key)]
+	return value, ok
+}
+
+func (cml *configMapAnnotations) UpdateAnnotation(key, value string) (err error) {
+	if cml.cm == nil {
+		return errors.New("endpoint not initialized, call Init first")
+	}
+	// speculatively change the state, then replace with whatever comes back
+	// from Update(), which will be the latest on the server whatever happened
+	cml.cm.Annotations[cleanKey(key)] = value
+	cml.cm, err = cml.Client.ConfigMaps(cml.Namespace).Update(cml.cm)
+	return err
+}
+
+func (cml *configMapAnnotations) RemoveAnnotation(key string) (err error) {
+	if cml.cm == nil {
+		return errors.New("endpoint not initialized, call Init first")
+	}
+	// speculatively change the state, then replace with whatever comes back
+	// from Update(), which will be the latest on the server whatever happened
+	delete(cml.cm.Annotations, cleanKey(key))
+	cml.cm, err = cml.Client.ConfigMaps(cml.Namespace).Update(cml.cm)
+	return err
+}
+
+func (cml *configMapAnnotations) RemoveAnnotationsWithValue(valueToRemove string) (err error) {
+	if cml.cm == nil {
+		return errors.New("endpoint not initialized, call Init first")
+	}
+	// speculatively change the state, then replace with whatever comes back
+	// from Update(), which will be the latest on the server whatever happened
+	for key, value := range cml.cm.Annotations {
+		if value == valueToRemove {
+			delete(cml.cm.Annotations, key) // don't need to clean this key as it came from the map
+		}
+	}
+	cml.cm, err = cml.Client.ConfigMaps(cml.Namespace).Update(cml.cm)
+	return err
+}
+
+// Loop with jitter, fetching the cml data and calling f() until it
+// doesn't get an optimistic locking conflict.
+// If it succeeds or gets any other kind of error, stop the loop.
+func (cml *configMapAnnotations) LoopUpdate(f func() error) error {
+	stop := make(chan struct{})
+	var err error
+	wait.JitterUntil(func() {
+		if err = cml.Init(); err != nil {
+			close(stop)
+			return
+		}
+		err = f()
+		if err != nil && kubeErrors.IsConflict(err) {
+			log.Printf("Optimistic locking conflict: trying again: %s", err)
+			return
+		}
+		close(stop)
+	}, retryPeriod, jitterFactor, true, stop)
+	return err
+}

prog/kube-peers/main.go

@@ -1,40 +1,36 @@
+/*
+Package main deals with weave-net peers on the cluster.
+
+This involves peer management, such as getting the latest peers or removing defunct peers from the cluster
+*/
 package main
 
 import (
+	"flag"
 	"fmt"
-	"log"
 	"net"
+	"os"
 
+	api "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/client-go/kubernetes"
-	api "k8s.io/client-go/pkg/api/v1"
 	"k8s.io/client-go/rest"
+
+	weaveapi "github.com/weaveworks/weave/api"
+	"github.com/weaveworks/weave/common"
 )
 
-func getKubePeers() ([]string, error) {
-	config, err := rest.InClusterConfig()
-	if err != nil {
-		return nil, err
-	}
-	c, err := kubernetes.NewForConfig(config)
-	if err != nil {
-		return nil, err
-	}
-	nodeList, err := c.Nodes().List(api.ListOptions{})
-	if err != nil {
-		// Fallback for cases (e.g. from kube-up.sh) where kube-proxy is not running on master
-		config.Host = "http://localhost:8080"
-		log.Print("error contacting APIServer: ", err, "; trying with fallback: ", config.Host)
-		c, err = kubernetes.NewForConfig(config)
-		if err != nil {
-			return nil, err
-		}
-		nodeList, err = c.Nodes().List(api.ListOptions{})
-	}
+type nodeInfo struct {
+	name string
+	addr string
+}
 
+// return the IP addresses of all nodes in the cluster
+func getKubePeers(c *kubernetes.Clientset) ([]nodeInfo, error) {
+	nodeList, err := c.CoreV1().Nodes().List(api.ListOptions{})
 	if err != nil {
 		return nil, err
 	}
-	addresses := make([]string, 0, len(nodeList.Items))
+	addresses := make([]nodeInfo, 0, len(nodeList.Items))
 	for _, peer := range nodeList.Items {
 		var internalIP, externalIP string
 		for _, addr := range peer.Status.Addresses {
@@ -52,20 +48,189 @@ func getKubePeers() ([]string, error) {
 
 		// Fallback for cases where a Node has an ExternalIP but no InternalIP
 		if internalIP != "" {
-			addresses = append(addresses, internalIP)
+			addresses = append(addresses, nodeInfo{name: peer.Name, addr: internalIP})
 		} else if externalIP != "" {
-			addresses = append(addresses, externalIP)
+			addresses = append(addresses, nodeInfo{name: peer.Name, addr: externalIP})
 		}
 	}
 	return addresses, nil
 }
 
-func main() {
-	peers, err := getKubePeers()
-	if err != nil {
-		log.Fatalf("Could not get peers: %v", err)
+const (
+	configMapName      = "weave-net"
+	configMapNamespace = "kube-system"
+)
+
+// update the list of all peers that have gone through this code path
+func addMyselfToPeerList(cml *configMapAnnotations, c *kubernetes.Clientset, peerName, name string) (*peerList, error) {
+	var list *peerList
+	err := cml.LoopUpdate(func() error {
+		var err error
+		list, err = cml.GetPeerList()
+		if err != nil {
+			return err
+		}
+		if !list.contains(peerName) {
+			list.add(peerName, name)
+			err = cml.UpdatePeerList(*list)
+			if err != nil {
+				return err
+			}
+		}
+		return nil
+	})
+	return list, err
+}
+
+func checkIamInPeerList(cml *configMapAnnotations, c *kubernetes.Clientset, peerName string) (bool, error) {
+	if err := cml.Init(); err != nil {
+		return false, err
 	}
-	for _, addr := range peers {
-		fmt.Println(addr)
+	list, err := cml.GetPeerList()
+	if err != nil {
+		return false, err
+	}
+	common.Log.Debugf("[kube-peers] Checking peer %q against list %v", peerName, list)
+	return list.contains(peerName), nil
+}
+
+// For each of those peers that is no longer listed as a node by
+// Kubernetes, remove it from Weave IPAM
+func reclaimRemovedPeers(weave *weaveapi.Client, cml *configMapAnnotations, nodes []nodeInfo, myPeerName string) error {
+	for {
+		if err := cml.Init(); err != nil {
+			return err
+		}
+		// 1. Compare peers stored in the peerList against all peers reported by k8s now.
+		storedPeerList, err := cml.GetPeerList()
+		if err != nil {
+			return err
+		}
+		peerMap := make(map[string]peerInfo, len(storedPeerList.Peers))
+		for _, peer := range storedPeerList.Peers {
+			peerMap[peer.NodeName] = peer
+		}
+		for _, node := range nodes {
+			delete(peerMap, node.name)
+		}
+		common.Log.Debugln("[kube-peers] Nodes that have disappeared:", peerMap)
+		if len(peerMap) == 0 {
+			break
+		}
+		// 2. Loop for each X in the first set and not in the second - we wish to remove X from our data structures
+		for _, peer := range peerMap {
+			common.Log.Debugln("[kube-peers] Preparing to remove disappeared peer", peer)
+			okToRemove := false
+			// 3. Check if there is an existing annotation with key X
+			if existingAnnotation, found := cml.GetAnnotation(KubePeersPrefix + peer.PeerName); found {
+				common.Log.Debugln("[kube-peers] Existing annotation", existingAnnotation)
+				// 4.   If annotation already contains my identity, ok;
+				if existingAnnotation == myPeerName {
+					okToRemove = true
+				}
+			} else {
+				// 5.   If non-existent, write an annotation with key X and contents "my identity"
+				common.Log.Debugln("[kube-peers] Noting I plan to remove ", peer.PeerName)
+				if err := cml.UpdateAnnotation(KubePeersPrefix+peer.PeerName, myPeerName); err == nil {
+					okToRemove = true
+				} else {
+					common.Log.Debugln("[kube-peers] error from UpdateAnnotation: ", err)
+				}
+			}
+			if okToRemove {
+				// 6.   If step 4 or 5 succeeded, rmpeer X
+				result, err := weave.RmPeer(peer.PeerName)
+				common.Log.Infof("[kube-peers] rmpeer of %s: %s", peer.PeerName, result)
+				if err != nil {
+					return err
+				}
+				cml.LoopUpdate(func() error {
+					// 7aa.   Remove any annotations Z* that have contents X
+					if err := cml.RemoveAnnotationsWithValue(peer.PeerName); err != nil {
+						return err
+					}
+					// 7a.    Remove X from peerList
+					storedPeerList.remove(peer.PeerName)
+					if err := cml.UpdatePeerList(*storedPeerList); err != nil {
+						return err
+					}
+					// 7b.    Remove annotation with key X
+					return cml.RemoveAnnotation(KubePeersPrefix + peer.PeerName)
+				})
+			}
+			// 8.   If step 5 failed due to optimistic lock conflict, stop: someone else is handling X
+
+			// Step 3-5 is to protect against two simultaneous rmpeers of X
+			// Step 4 is to pick up again after a restart between step 5 and step 7b
+			// If the peer doing the reclaim disappears between steps 5 and 7a, then someone will clean it up in step 7aa
+			// If peer doing the reclaim disappears forever between 7a and 7b then we get a dangling annotation
+			// This should be sufficiently rare that we don't care.
+		}
+
+		// 9. Go back to step 1 until there is no difference between the two sets
+	}
+	// Question: Should we narrow step 2 by checking against Weave Net IPAM?
+	// i.e. If peer X owns any address space and is marked unreachable, we want to rmpeer X
+	return nil
+}
+
+func main() {
+	var (
+		justReclaim bool
+		justCheck   bool
+		peerName    string
+		nodeName    string
+		logLevel    string
+	)
+	flag.BoolVar(&justReclaim, "reclaim", false, "reclaim IP space from dead peers")
+	flag.BoolVar(&justCheck, "check-peer-new", false, "return success if peer name is not stored in annotation")
+	flag.StringVar(&peerName, "peer-name", "unknown", "name of this Weave Net peer")
+	flag.StringVar(&nodeName, "node-name", "unknown", "name of this Kubernetes node")
+	flag.StringVar(&logLevel, "log-level", "info", "logging level (debug, info, warning, error)")
+	flag.Parse()
+
+	common.SetLogLevel(logLevel)
+	config, err := rest.InClusterConfig()
+	if err != nil {
+		common.Log.Fatalf("[kube-peers] Could not get cluster config: %v", err)
+	}
+	c, err := kubernetes.NewForConfig(config)
+	if err != nil {
+		common.Log.Fatalf("[kube-peers] Could not make Kubernetes connection: %v", err)
+	}
+	if justCheck {
+		cml := newConfigMapAnnotations(configMapNamespace, configMapName, c)
+		exists, err := checkIamInPeerList(cml, c, peerName)
+		if err != nil {
+			common.Log.Fatalf("[kube-peers] Could not check peer list: %v", err)
+		}
+		if exists {
+			os.Exit(9)
+		} else {
+			os.Exit(0)
+		}
+	}
+	peers, err := getKubePeers(c)
+	if err != nil {
+		common.Log.Fatalf("[kube-peers] Could not get peers: %v", err)
+	}
+	if justReclaim {
+		cml := newConfigMapAnnotations(configMapNamespace, configMapName, c)
+
+		list, err := addMyselfToPeerList(cml, c, peerName, nodeName)
+		if err != nil {
+			common.Log.Fatalf("[kube-peers] Could not update peer list: %v", err)
+		}
+		common.Log.Infoln("[kube-peers] Added myself to peer list", list)
+
+		weave := weaveapi.NewClient(os.Getenv("WEAVE_HTTP_ADDR"), common.Log)
+		err = reclaimRemovedPeers(weave, cml, peers, peerName)
+		if err != nil {
+			common.Log.Fatalf("[kube-peers] Error while reclaiming space: %v", err)
+		}
+		return
+	}
+	for _, node := range peers {
+		fmt.Println(node.addr)
 	}
 }

prog/net-plugin/config.json

@@ -0,0 +1,98 @@
+{
+  "description": "Weave Net plugin for Docker",
+  "documentation": "https://weave.works",
+  "entrypoint": [
+    "/home/weave/launch.sh"
+  ],
+  "interface": {
+    "socket": "weave.sock",
+    "types": [
+      "docker.networkdriver/1.0"
+    ]
+  },
+  "linux": {
+    "capabilities": [
+      "CAP_SYS_ADMIN",
+      "CAP_NET_ADMIN",
+      "CAP_SYS_MODULE"
+    ]
+  },
+  "mounts": [
+    {
+      "destination": "/host/proc/",
+      "type": "bind",
+      "source": "/proc/",
+      "options": ["rbind", "rw"]
+    },
+    {
+      "destination": "/var/run/docker.sock",
+      "source": "/var/run/docker.sock",
+      "type": "bind",
+      "options": ["rbind"]
+    },
+    {
+      "destination": "/host/var/lib/",
+      "source": "/var/lib/",
+      "type": "bind",
+      "options": ["rbind"]
+    },
+    {
+      "destination": "/host/etc/",
+      "source": "/etc/",
+      "type": "bind",
+      "options": ["rbind"]
+    },
+    {
+      "destination": "/lib/modules/",
+      "source": "/lib/modules/",
+      "type": "bind",
+      "options": ["rbind"]
+    }
+  ],
+  "env": [
+    {
+      "description": "Extra args to `weaver` and `plugin`",
+      "name": "EXTRA_ARGS",
+      "settable": [
+          "value"
+      ],
+      "value": ""
+    },
+    {
+      "description": "Encryption password",
+      "name": "WEAVE_PASSWORD",
+      "settable": [
+          "value"
+      ],
+      "value": ""
+    },
+    {
+      "description": "MTU",
+      "name": "WEAVE_MTU",
+      "settable": [
+          "value"
+      ],
+      "value": ""
+    },
+    {
+      "description": "Enable multicast for all Weave networks",
+      "name": "WEAVE_MULTICAST",
+      "settable": [
+          "value"
+      ],
+      "value": ""
+    },
+    {
+      "description": "The range of IP addresses used by Weave Net",
+      "name": "IPALLOC_RANGE",
+      "settable": [
+          "value"
+      ],
+      "value": ""
+    }
+  ],
+  "network": {
+    "type": "host"
+  },
+  "workdir": ""
+}

prog/net-plugin/launch.sh

@@ -0,0 +1,53 @@
+#!/bin/sh
+
+set -e
+
+# Default if not supplied - same as weave net default
+IPALLOC_RANGE=${IPALLOC_RANGE:-10.32.0.0/12}
+HTTP_ADDR=${WEAVE_HTTP_ADDR:-127.0.0.1:6784}
+STATUS_ADDR=${WEAVE_STATUS_ADDR:-0.0.0.0:6782}
+HOST_ROOT=${HOST_ROOT:-/host}
+WEAVE_DIR="/host/var/lib/weave"
+
+mkdir $WEAVE_DIR || true
+
+echo "Starting launch.sh"
+
+# Check if the IP range overlaps anything existing on the host
+/usr/bin/weaveutil netcheck $IPALLOC_RANGE weave
+
+# We need to get a list of Swarm nodes which might run the net-plugin:
+# - In the case of missing restart.sentinel, we assume that net-plugin has started
+#   for the first time via the docker-plugin cmd. So it's safe to request docker.sock.
+# - If restart.sentinel present, let weaver restore from it as docker.sock is not
+#   available to any plugin in general (https://github.com/moby/moby/issues/32815).
+
+PEERS=
+if [ ! -f "/restart.sentinel" ]; then
+    PEERS=$(/usr/bin/weaveutil swarm-manager-peers)
+fi
+
+router_bridge_opts() {
+    echo --datapath=datapath
+    [ -z "$WEAVE_MTU" ] || echo --mtu "$WEAVE_MTU"
+    [ -z "$WEAVE_NO_FASTDP" ] || echo --no-fastdp
+}
+
+multicast_opt() {
+    [ -z "$WEAVE_MULTICAST" ] || echo "--plugin-v2-multicast"
+}
+
+exec /home/weave/weaver $EXTRA_ARGS --port=6783 $(router_bridge_opts) \
+    --host-root=/host \
+    --proc-path=/host/proc \
+    --http-addr=$HTTP_ADDR --status-addr=$STATUS_ADDR \
+    --no-dns \
+    --ipalloc-range=$IPALLOC_RANGE \
+    --nickname "$(hostname)" \
+    --log-level=debug \
+    --db-prefix="$WEAVE_DIR/weave" \
+    --plugin-v2 \
+    $(multicast_opt) \
+    --plugin-mesh-socket='' \
+    --docker-api='' \
+    $PEERS

prog/plugin/Dockerfile.template

@@ -1,3 +0,0 @@
-FROM DOCKERHUB_USER/weaveARCH_EXT
-ADD ./plugin /home/weave/
-ENTRYPOINT ["/home/weave/plugin"]

prog/plugin/main.go

@@ -1,140 +0,0 @@
-package main
-
-import (
-	"flag"
-	"fmt"
-	"net"
-	"net/http"
-	"os"
-	"os/signal"
-	"path"
-	"strings"
-	"syscall"
-
-	"github.com/docker/libnetwork/ipamapi"
-	weaveapi "github.com/weaveworks/weave/api"
-	"github.com/weaveworks/weave/common"
-	"github.com/weaveworks/weave/common/docker"
-	weavenet "github.com/weaveworks/weave/net"
-	ipamplugin "github.com/weaveworks/weave/plugin/ipam"
-	netplugin "github.com/weaveworks/weave/plugin/net"
-	"github.com/weaveworks/weave/plugin/skel"
-)
-
-var version = "unreleased"
-
-var Log = common.Log
-
-func main() {
-	var (
-		justVersion      bool
-		address          string
-		meshAddress      string
-		logLevel         string
-		noMulticastRoute bool
-	)
-
-	flag.BoolVar(&justVersion, "version", false, "print version and exit")
-	flag.StringVar(&logLevel, "log-level", "info", "logging level (debug, info, warning, error)")
-	flag.StringVar(&address, "socket", "/run/docker/plugins/weave.sock", "socket on which to listen")
-	flag.StringVar(&meshAddress, "meshsocket", "/run/docker/plugins/weavemesh.sock", "socket on which to listen in mesh mode")
-	flag.BoolVar(&noMulticastRoute, "no-multicast-route", false, "deprecated (this is now the default)")
-
-	flag.Parse()
-
-	if justVersion {
-		fmt.Printf("weave plugin %s\n", version)
-		os.Exit(0)
-	}
-
-	common.SetLogLevel(logLevel)
-
-	weave := weaveapi.NewClient(os.Getenv("WEAVE_HTTP_ADDR"), Log)
-
-	// API 1.21 is the first version that supports docker network commands
-	dockerClient, err := docker.NewVersionedClientFromEnv("1.21")
-	if err != nil {
-		Log.Fatalf("unable to connect to docker: %s", err)
-	}
-
-	Log.Println("Weave plugin", version, "Command line options:", os.Args[1:])
-	if noMulticastRoute {
-		Log.Warning("--no-multicast-route option has been removed; multicast is off by default")
-	}
-	Log.Info(dockerClient.Info())
-
-	err = run(dockerClient, weave, address, meshAddress)
-	if err != nil {
-		Log.Fatal(err)
-	}
-}
-
-func run(dockerClient *docker.Client, weave *weaveapi.Client, address, meshAddress string) error {
-	endChan := make(chan error, 1)
-	if address != "" {
-		globalListener, err := listenAndServe(dockerClient, weave, address, endChan, "global", false)
-		if err != nil {
-			return err
-		}
-		defer os.Remove(address)
-		defer globalListener.Close()
-	}
-	if meshAddress != "" {
-		meshListener, err := listenAndServe(dockerClient, weave, meshAddress, endChan, "local", true)
-		if err != nil {
-			return err
-		}
-		defer os.Remove(meshAddress)
-		defer meshListener.Close()
-	}
-
-	statusListener, err := weavenet.ListenUnixSocket("/home/weave/status.sock")
-	if err != nil {
-		return err
-	}
-	go serveStatus(statusListener)
-	sigChan := make(chan os.Signal, 1)
-	signal.Notify(sigChan, os.Interrupt, syscall.SIGTERM)
-
-	select {
-	case sig := <-sigChan:
-		Log.Debugf("Caught signal %s; shutting down", sig)
-		return nil
-	case err := <-endChan:
-		return err
-	}
-}
-
-func listenAndServe(dockerClient *docker.Client, weave *weaveapi.Client, address string, endChan chan<- error, scope string, withIpam bool) (net.Listener, error) {
-	name := strings.TrimSuffix(path.Base(address), ".sock")
-	d, err := netplugin.New(dockerClient, weave, name, scope)
-	if err != nil {
-		return nil, err
-	}
-
-	var i ipamapi.Ipam
-	if withIpam {
-		i = ipamplugin.NewIpam(weave)
-	}
-
-	listener, err := weavenet.ListenUnixSocket(address)
-	if err != nil {
-		return nil, err
-	}
-	Log.Printf("Listening on %s for %s scope", address, scope)
-
-	go func() {
-		endChan <- skel.Listen(listener, d, i)
-	}()
-
-	return listener, nil
-}
-
-func serveStatus(listener net.Listener) {
-	server := &http.Server{Handler: http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
-		fmt.Fprintln(w, "ok")
-	})}
-	if err := server.Serve(listener); err != nil {
-		Log.Fatalf("ListenAndServeStatus failed: %s", err)
-	}
-}

prog/plugin/main_test.go

@@ -1,12 +0,0 @@
-package main
-
-import (
-	"testing"
-
-	weavetest "github.com/weaveworks/weave/testing"
-)
-
-func TestMain(t *testing.T) {
-	weavetest.TrimTestArgs()
-	main()
-}

prog/weave-kube/launch.sh

@@ -30,6 +30,7 @@ HTTP_ADDR=${WEAVE_HTTP_ADDR:-127.0.0.1:6784}
 STATUS_ADDR=${WEAVE_STATUS_ADDR:-0.0.0.0:6782}
 HOST_ROOT=${HOST_ROOT:-/host}
 CONN_LIMIT=${CONN_LIMIT:-30}
+DB_PREFIX=${DB_PREFIX:-/weavedb/weave-net}
 
 # Check if the IP range overlaps anything existing on the host
 /usr/bin/weaveutil netcheck $IPALLOC_RANGE weave
@@ -53,15 +54,12 @@ else
     fi
 fi
 
-# Explicitly create the bridge so we can pass --expect-npc
+# Default is that npc will be running; allow caller to override
 WEAVE_NPC_OPTS="--expect-npc"
 if [ "${EXPECT_NPC}" = "0" ]; then
     WEAVE_NPC_OPTS=""
 fi
 
-export HOST_ROOT=/host
-/home/weave/weave --local create-bridge --force $WEAVE_NPC_OPTS
-
 # Kubernetes sets HOSTNAME to the host's hostname
 # when running a pod in host namespace.
 NICKNAME_ARG=""
@@ -69,17 +67,11 @@ if [ -n "$HOSTNAME" ] ; then
     NICKNAME_ARG="--nickname=$HOSTNAME"
 fi
 
-BRIDGE_OPTIONS="--datapath=datapath"
-if [ "$(/home/weave/weave --local bridge-type)" = "bridge" ] ; then
-    # TODO: Call into weave script to do this
-    if ! ip link show vethwe-pcap >/dev/null 2>&1 ; then
-        ip link add name vethwe-bridge type veth peer name vethwe-pcap
-        ip link set vethwe-bridge up
-        ip link set vethwe-pcap up
-        ip link set vethwe-bridge master weave
-    fi
-    BRIDGE_OPTIONS="--iface=vethwe-pcap"
-fi
+router_bridge_opts() {
+    echo --datapath=datapath
+    [ -z "$WEAVE_MTU" ] || echo --mtu "$WEAVE_MTU"
+    [ -z "$WEAVE_NO_FASTDP" ] || echo --no-fastdp
+}
 
 if [ -z "$KUBE_PEERS" ]; then
     if ! KUBE_PEERS=$(/home/weave/kube-peers) || [ -z "$KUBE_PEERS" ]; then
@@ -96,6 +88,29 @@ if [ -z "$IPALLOC_INIT" ]; then
     IPALLOC_INIT="consensus=$(peer_count $KUBE_PEERS)"
 fi
 
+# Find out what peer name we will be using
+PEERNAME=$(/home/weave/weaver $EXTRA_ARGS --print-peer-name --host-root=$HOST_ROOT --db-prefix="$DB_PREFIX")
+if [ -z "$PEERNAME" ]; then
+    echo "Unable to get peer name" >&2
+    exit 2
+fi
+
+# If we have persisted data and peer reclaim is in use
+if [ -f ${DB_PREFIX}data.db ]; then
+    if /usr/bin/weaveutil get-db-flag "$DB_PREFIX" peer-reclaim >/dev/null ; then
+        # If this peer name is not stored in the list then we were removed by another peer.
+        if /home/weave/kube-peers -check-peer-new -peer-name="$PEERNAME" -log-level=debug ; then
+            # In order to avoid a CRDT clash, clean up
+            echo "Peer not in list; removing persisted data" >&2
+            rm -f ${DB_PREFIX}data.db
+        fi
+    fi
+fi
+
+# flag that peer reclaim is now in use - note we have to do this before
+# weaver is running as the DB can only be opened by one process at a time
+/usr/bin/weaveutil set-db-flag "$DB_PREFIX" peer-reclaim ok
+
 post_start_actions() {
     # Wait for weave process to become responsive
     while true ; do
@@ -114,18 +129,26 @@ post_start_actions() {
         fi
     fi
     mkdir -p $HOST_ROOT/etc/cni/net.d
+    export HOST_ROOT
     /home/weave/weave --local setup-cni
 
+    # Attempt to run the reclaim process, but don't halt the script if it fails
+    /home/weave/kube-peers -reclaim -node-name="$HOSTNAME" -peer-name="$PEERNAME" -log-level=debug || true
+
     # Expose the weave network so host processes can communicate with pods
     /home/weave/weave --local expose $WEAVE_EXPOSE_IP
 }
 
 post_start_actions &
 
-/home/weave/weaver $EXTRA_ARGS --port=6783 $BRIDGE_OPTIONS \
+/home/weave/weaver $EXTRA_ARGS --port=6783 $(router_bridge_opts) \
+     --name="$PEERNAME" \
+     --host-root=$HOST_ROOT \
      --http-addr=$HTTP_ADDR --status-addr=$STATUS_ADDR --docker-api='' --no-dns \
+     --db-prefix="$DB_PREFIX" \
      --ipalloc-range=$IPALLOC_RANGE $NICKNAME_ARG \
      --ipalloc-init $IPALLOC_INIT \
      --conn-limit=$CONN_LIMIT \
+     $WEAVE_NPC_OPTS \
      "$@" \
      $KUBE_PEERS

prog/weave-kube/weave-daemonset-k8s-1.6.yaml

@@ -1,120 +1,180 @@
-kind: ClusterRole
-apiVersion: rbac.authorization.k8s.io/v1beta1
-metadata:
-  name: weave-net
-rules:
-- apiGroups:
-  - ""
-  resources:
-  - pods
-  - namespaces
-  - nodes
-  verbs:
-  - get
-  - list
-  - watch
-- apiGroups:
-  - extensions
-  resources:
-  - networkpolicies
-  verbs:
-  - get
-  - list
-  - watch
----
 apiVersion: v1
-kind: ServiceAccount
-metadata:
-  name: weave-net
-  namespace: kube-system
----
-kind: ClusterRoleBinding
-apiVersion: rbac.authorization.k8s.io/v1beta1
-metadata:
-  name: weave-net
-roleRef:
-  apiGroup: rbac.authorization.k8s.io
-  kind: ClusterRole
-  name: weave-net
-subjects:
-- kind: ServiceAccount
-  name: weave-net
-  namespace: kube-system
----
-apiVersion: extensions/v1beta1
-kind: DaemonSet
-metadata:
-  name: weave-net
-  namespace: kube-system
-spec:
-  template:
+kind: List
+items:
+  - apiVersion: v1
+    kind: ServiceAccount
     metadata:
+      name: weave-net
       labels:
         name: weave-net
+      namespace: kube-system
+  - apiVersion: rbac.authorization.k8s.io/v1beta1
+    kind: ClusterRole
+    metadata:
+      name: weave-net
+      labels:
+        name: weave-net
+    rules:
+      - apiGroups:
+          - ''
+        resources:
+          - pods
+          - namespaces
+          - nodes
+        verbs:
+          - get
+          - list
+          - watch
+      - apiGroups:
+          - extensions
+        resources:
+          - networkpolicies
+        verbs:
+          - get
+          - list
+          - watch
+  - apiVersion: rbac.authorization.k8s.io/v1beta1
+    kind: ClusterRoleBinding
+    metadata:
+      name: weave-net
+      labels:
+        name: weave-net
+    roleRef:
+      kind: ClusterRole
+      name: weave-net
+      apiGroup: rbac.authorization.k8s.io
+    subjects:
+      - kind: ServiceAccount
+        name: weave-net
+        namespace: kube-system
+  - apiVersion: rbac.authorization.k8s.io/v1beta1
+    kind: Role
+    metadata:
+      name: weave-net-kube-peer
+      namespace: kube-system
+      labels:
+        name: weave-net-kube-peer
+    rules:
+      - apiGroups:
+          - ''
+        resources:
+          - configmaps
+        resourceNames:
+          - weave-net
+        verbs:
+          - get
+          - update
+      - apiGroups:
+          - ''
+        resources:
+          - configmaps
+        verbs:
+          - create
+  - apiVersion: rbac.authorization.k8s.io/v1beta1
+    kind: RoleBinding
+    metadata:
+      name: weave-net-kube-peer
+      namespace: kube-system
+      labels:
+        name: weave-net-kube-peer
+    roleRef:
+      kind: Role
+      name: weave-net-kube-peer
+      apiGroup: rbac.authorization.k8s.io
+    subjects:
+      - kind: ServiceAccount
+        name: weave-net
+        namespace: kube-system
+  - apiVersion: extensions/v1beta1
+    kind: DaemonSet
+    metadata:
+      name: weave-net
+      labels:
+        name: weave-net
+      namespace: kube-system
     spec:
-      hostNetwork: true
-      hostPID: true
-      containers:
-        - name: weave
-          image: weaveworks/weave-kube:latest
-          imagePullPolicy: Always
-          command:
-            - /home/weave/launch.sh
-          livenessProbe:
-            initialDelaySeconds: 30
-            httpGet:
-              host: 127.0.0.1
-              path: /status
-              port: 6784
+      template:
+        metadata:
+          labels:
+            name: weave-net
+        spec:
+          containers:
+            - name: weave
+              command:
+                - /home/weave/launch.sh
+              env:
+                - name: HOSTNAME
+                  valueFrom:
+                    fieldRef:
+                      apiVersion: v1
+                      fieldPath: spec.nodeName
+              image: 'weaveworks/weave-kube:latest'
+              imagePullPolicy: Always
+              livenessProbe:
+                httpGet:
+                  host: 127.0.0.1
+                  path: /status
+                  port: 6784
+                initialDelaySeconds: 30
+              resources:
+                requests:
+                  cpu: 10m
+              securityContext:
+                privileged: true
+              volumeMounts:
+                - name: weavedb
+                  mountPath: /weavedb
+                - name: cni-bin
+                  mountPath: /host/opt
+                - name: cni-bin2
+                  mountPath: /host/home
+                - name: cni-conf
+                  mountPath: /host/etc
+                - name: dbus
+                  mountPath: /host/var/lib/dbus
+                - name: lib-modules
+                  mountPath: /lib/modules
+            - name: weave-npc
+              env:
+                - name: HOSTNAME
+                  valueFrom:
+                    fieldRef:
+                      apiVersion: v1
+                      fieldPath: spec.nodeName
+              image: 'weaveworks/weave-npc:latest'
+              imagePullPolicy: Always
+              resources:
+                requests:
+                  cpu: 10m
+              securityContext:
+                privileged: true
+          hostNetwork: true
+          hostPID: true
+          restartPolicy: Always
           securityContext:
-            privileged: true
-          volumeMounts:
+            seLinuxOptions: {}
+          serviceAccountName: weave-net
+          tolerations:
+            - effect: NoSchedule
+              operator: Exists
+          volumes:
             - name: weavedb
-              mountPath: /weavedb
+              hostPath:
+                path: /var/lib/weave
             - name: cni-bin
-              mountPath: /host/opt
+              hostPath:
+                path: /opt
             - name: cni-bin2
-              mountPath: /host/home
+              hostPath:
+                path: /home
             - name: cni-conf
-              mountPath: /host/etc
+              hostPath:
+                path: /etc
             - name: dbus
-              mountPath: /host/var/lib/dbus
+              hostPath:
+                path: /var/lib/dbus
             - name: lib-modules
-              mountPath: /lib/modules
-          resources:
-            requests:
-              cpu: 10m
-        - name: weave-npc
-          image: weaveworks/weave-npc:latest
-          imagePullPolicy: Always
-          resources:
-            requests:
-              cpu: 10m
-          securityContext:
-            privileged: true
-      restartPolicy: Always
-      tolerations:
-      - key: node-role.kubernetes.io/master
-        effect: NoSchedule
-      serviceAccountName: weave-net
-      securityContext:
-        seLinuxOptions:
-          type: spc_t
-      volumes:
-        - name: weavedb
-          emptyDir: {}
-        - name: cni-bin
-          hostPath:
-            path: /opt
-        - name: cni-bin2
-          hostPath:
-            path: /home
-        - name: cni-conf
-          hostPath:
-            path: /etc
-        - name: dbus
-          hostPath:
-            path: /var/lib/dbus
-        - name: lib-modules
-          hostPath:
-            path: /lib/modules
+              hostPath:
+                path: /lib/modules
+      updateStrategy:
+        type: RollingUpdate

prog/weave-kube/weave-daemonset-k8s-1.7.yaml

@@ -0,0 +1,199 @@
+apiVersion: v1
+kind: List
+items:
+  - apiVersion: v1
+    kind: ServiceAccount
+    metadata:
+      name: weave-net
+      labels:
+        name: weave-net
+      namespace: kube-system
+  - apiVersion: rbac.authorization.k8s.io/v1beta1
+    kind: ClusterRole
+    metadata:
+      name: weave-net
+      labels:
+        name: weave-net
+    rules:
+      - apiGroups:
+          - ''
+        resources:
+          - pods
+          - namespaces
+          - nodes
+        verbs:
+          - get
+          - list
+          - watch
+      - apiGroups:
+          - extensions
+        resources:
+          - networkpolicies
+        verbs:
+          - get
+          - list
+          - watch
+      - apiGroups:
+          - 'networking.k8s.io'
+        resources:
+          - networkpolicies
+        verbs:
+          - get
+          - list
+          - watch
+  - apiVersion: rbac.authorization.k8s.io/v1beta1
+    kind: ClusterRoleBinding
+    metadata:
+      name: weave-net
+      labels:
+        name: weave-net
+    roleRef:
+      kind: ClusterRole
+      name: weave-net
+      apiGroup: rbac.authorization.k8s.io
+    subjects:
+      - kind: ServiceAccount
+        name: weave-net
+        namespace: kube-system
+  - apiVersion: rbac.authorization.k8s.io/v1beta1
+    kind: Role
+    metadata:
+      name: weave-net
+      namespace: kube-system
+      labels:
+        name: weave-net
+    rules:
+      - apiGroups:
+          - ''
+        resources:
+          - configmaps
+        resourceNames:
+          - weave-net
+        verbs:
+          - get
+          - update
+      - apiGroups:
+          - ''
+        resources:
+          - configmaps
+        verbs:
+          - create
+  - apiVersion: rbac.authorization.k8s.io/v1beta1
+    kind: RoleBinding
+    metadata:
+      name: weave-net
+      namespace: kube-system
+      labels:
+        name: weave-net
+    roleRef:
+      kind: Role
+      name: weave-net
+      apiGroup: rbac.authorization.k8s.io
+    subjects:
+      - kind: ServiceAccount
+        name: weave-net
+        namespace: kube-system
+  - apiVersion: extensions/v1beta1
+    kind: DaemonSet
+    metadata:
+      name: weave-net
+      labels:
+        name: weave-net
+      namespace: kube-system
+    spec:
+      template:
+        metadata:
+          labels:
+            name: weave-net
+        spec:
+          containers:
+            - name: weave
+              command:
+                - /home/weave/launch.sh
+              env:
+                - name: HOSTNAME
+                  valueFrom:
+                    fieldRef:
+                      apiVersion: v1
+                      fieldPath: spec.nodeName
+              image: 'weaveworks/weave-kube:latest'
+              imagePullPolicy: Always
+              livenessProbe:
+                httpGet:
+                  host: 127.0.0.1
+                  path: /status
+                  port: 6784
+                initialDelaySeconds: 30
+              resources:
+                requests:
+                  cpu: 10m
+              securityContext:
+                privileged: true
+              volumeMounts:
+                - name: weavedb
+                  mountPath: /weavedb
+                - name: cni-bin
+                  mountPath: /host/opt
+                - name: cni-bin2
+                  mountPath: /host/home
+                - name: cni-conf
+                  mountPath: /host/etc
+                - name: dbus
+                  mountPath: /host/var/lib/dbus
+                - name: lib-modules
+                  mountPath: /lib/modules
+                - name: xtables-lock
+                  mountPath: /run/xtables.lock
+                  readOnly: false
+            - name: weave-npc
+              env:
+                - name: HOSTNAME
+                  valueFrom:
+                    fieldRef:
+                      apiVersion: v1
+                      fieldPath: spec.nodeName
+              image: 'weaveworks/weave-npc:latest'
+              imagePullPolicy: Always
+#npc-args
+              resources:
+                requests:
+                  cpu: 10m
+              securityContext:
+                privileged: true
+              volumeMounts:
+                - name: xtables-lock
+                  mountPath: /run/xtables.lock
+                  readOnly: false
+          hostNetwork: true
+          hostPID: true
+          restartPolicy: Always
+          securityContext:
+            seLinuxOptions: {}
+          serviceAccountName: weave-net
+          tolerations:
+            - effect: NoSchedule
+              operator: Exists
+          volumes:
+            - name: weavedb
+              hostPath:
+                path: /var/lib/weave
+            - name: cni-bin
+              hostPath:
+                path: /opt
+            - name: cni-bin2
+              hostPath:
+                path: /home
+            - name: cni-conf
+              hostPath:
+                path: /etc
+            - name: dbus
+              hostPath:
+                path: /var/lib/dbus
+            - name: lib-modules
+              hostPath:
+                path: /lib/modules
+            - name: xtables-lock
+              hostPath:
+                path: /run/xtables.lock
+      updateStrategy:
+        type: RollingUpdate

prog/weave-kube/weave-daemonset.yaml

@@ -1,80 +1,101 @@
-apiVersion: extensions/v1beta1
-kind: DaemonSet
-metadata:
-  name: weave-net
-  namespace: kube-system
-spec:
-  template:
+apiVersion: v1
+kind: List
+items:
+  - apiVersion: v1
+    kind: ServiceAccount
     metadata:
+      name: weave-net
       labels:
         name: weave-net
-      annotations:
-        scheduler.alpha.kubernetes.io/tolerations: |
-          [
-            {
-              "key": "dedicated",
-              "operator": "Equal",
-              "value": "master",
-              "effect": "NoSchedule"
-            }
-          ]
+      namespace: kube-system
+  - apiVersion: extensions/v1beta1
+    kind: DaemonSet
+    metadata:
+      name: weave-net
+      labels:
+        name: weave-net
+      namespace: kube-system
     spec:
-      hostNetwork: true
-      hostPID: true
-      containers:
-        - name: weave
-          image: weaveworks/weave-kube:latest
-          imagePullPolicy: Always
-          command:
-            - /home/weave/launch.sh
-          livenessProbe:
-            initialDelaySeconds: 30
-            httpGet:
-              host: 127.0.0.1
-              path: /status
-              port: 6784
+      template:
+        metadata:
+          annotations:
+            scheduler.alpha.kubernetes.io/tolerations: >-
+              [{"key":"dedicated","operator":"Equal","value":"master","effect":"NoSchedule"}]
+          labels:
+            name: weave-net
+        spec:
+          containers:
+            - name: weave
+              command:
+                - /home/weave/launch.sh
+              env:
+                - name: HOSTNAME
+                  valueFrom:
+                    fieldRef:
+                      apiVersion: v1
+                      fieldPath: spec.nodeName
+              image: 'weaveworks/weave-kube:latest'
+              imagePullPolicy: Always
+              livenessProbe:
+                httpGet:
+                  host: 127.0.0.1
+                  path: /status
+                  port: 6784
+                initialDelaySeconds: 30
+              resources:
+                requests:
+                  cpu: 10m
+              securityContext:
+                privileged: true
+              volumeMounts:
+                - name: weavedb
+                  mountPath: /weavedb
+                - name: cni-bin
+                  mountPath: /host/opt
+                - name: cni-bin2
+                  mountPath: /host/home
+                - name: cni-conf
+                  mountPath: /host/etc
+                - name: dbus
+                  mountPath: /host/var/lib/dbus
+                - name: lib-modules
+                  mountPath: /lib/modules
+            - name: weave-npc
+              env:
+                - name: HOSTNAME
+                  valueFrom:
+                    fieldRef:
+                      apiVersion: v1
+                      fieldPath: spec.nodeName
+              image: 'weaveworks/weave-npc:latest'
+              imagePullPolicy: Always
+              resources:
+                requests:
+                  cpu: 10m
+              securityContext:
+                privileged: true
+          hostNetwork: true
+          hostPID: true
+          restartPolicy: Always
           securityContext:
-            privileged: true
-          volumeMounts:
+            seLinuxOptions: {}
+          serviceAccountName: weave-net
+          volumes:
             - name: weavedb
-              mountPath: /weavedb
+              hostPath:
+                path: /var/lib/weave
             - name: cni-bin
-              mountPath: /host/opt
+              hostPath:
+                path: /opt
             - name: cni-bin2
-              mountPath: /host/home
+              hostPath:
+                path: /home
             - name: cni-conf
-              mountPath: /host/etc
+              hostPath:
+                path: /etc
             - name: dbus
-              mountPath: /host/var/lib/dbus
+              hostPath:
+                path: /var/lib/dbus
             - name: lib-modules
-              mountPath: /lib/modules
-          resources:
-            requests:
-              cpu: 10m
-        - name: weave-npc
-          image: weaveworks/weave-npc:latest
-          imagePullPolicy: Always
-          resources:
-            requests:
-              cpu: 10m
-          securityContext:
-            privileged: true
-      restartPolicy: Always
-      volumes:
-        - name: weavedb
-          emptyDir: {}
-        - name: cni-bin
-          hostPath:
-            path: /opt
-        - name: cni-bin2
-          hostPath:
-            path: /home
-        - name: cni-conf
-          hostPath:
-            path: /etc
-        - name: dbus
-          hostPath:
-            path: /var/lib/dbus
-        - name: lib-modules
-          hostPath:
-            path: /lib/modules
+              hostPath:
+                path: /lib/modules

prog/weave-npc/main.go

@@ -7,12 +7,13 @@ import (
 
 	"github.com/coreos/go-iptables/iptables"
 	"github.com/spf13/cobra"
+	coreapi "k8s.io/api/core/v1"
+	extnapi "k8s.io/api/extensions/v1beta1"
+	networkingv1 "k8s.io/api/networking/v1"
+	"k8s.io/apimachinery/pkg/fields"
+	"k8s.io/apimachinery/pkg/runtime"
+	"k8s.io/apimachinery/pkg/util/wait"
 	"k8s.io/client-go/kubernetes"
-	coreapi "k8s.io/client-go/pkg/api/v1"
-	extnapi "k8s.io/client-go/pkg/apis/extensions/v1beta1"
-	"k8s.io/client-go/pkg/fields"
-	"k8s.io/client-go/pkg/runtime"
-	"k8s.io/client-go/pkg/util/wait"
 	"k8s.io/client-go/rest"
 	"k8s.io/client-go/tools/cache"
 
@@ -29,12 +30,13 @@ var (
 	logLevel    string
 	allowMcast  bool
 	nodeName    string
+	legacy      bool
 )
 
 func handleError(err error) { common.CheckFatal(err) }
 
 func makeController(getter cache.Getter, resource string,
-	objType runtime.Object, handlers cache.ResourceEventHandlerFuncs) *cache.Controller {
+	objType runtime.Object, handlers cache.ResourceEventHandlerFuncs) cache.Controller {
 	listWatch := cache.NewListWatchFromClient(getter, resource, "", fields.Everything())
 	_, controller := cache.NewInformer(listWatch, objType, 0, handlers)
 	return controller
@@ -50,11 +52,7 @@ func resetIPTables(ipt *iptables.IPTables) error {
 		return err
 	}
 
-	if err := ipt.ClearChain(npc.TableFilter, npc.MainChain); err != nil {
-		return err
-	}
-
-	return nil
+	return ipt.ClearChain(npc.TableFilter, npc.MainChain)
 }
 
 func resetIPSets(ips ipset.Interface) error {
@@ -116,16 +114,13 @@ func createBaseRules(ipt *iptables.IPTables, ips ipset.Interface) error {
 	if err := ips.Create(npc.LocalIpset, ipset.HashIP); err != nil {
 		return err
 	}
-	if err := ipt.Append(npc.TableFilter, npc.MainChain,
-		"-m", "set", "--match-set", npc.LocalIpset, "src",
-		"-m", "set", "!", "--match-set", npc.LocalIpset, "dst", "-j", "ACCEPT"); err != nil {
-		return err
-	}
-
-	return nil
+	return ipt.Append(npc.TableFilter, npc.MainChain,
+		"-m", "set", "!", "--match-set", npc.LocalIpset, "dst", "-j", "ACCEPT")
 }
 
 func root(cmd *cobra.Command, args []string) {
+	var npController cache.Controller
+
 	common.SetLogLevel(logLevel)
 	if nodeName == "" {
 		// HOSTNAME is set by Kubernetes for pods in the host network namespace
@@ -136,6 +131,10 @@ func root(cmd *cobra.Command, args []string) {
 	}
 	common.Log.Infof("Starting Weaveworks NPC %s; node name %q", version, nodeName)
 
+	if legacy {
+		common.Log.Info("Running in legacy mode (k8s pre-1.7 network policy semantics)")
+	}
+
 	if err := metrics.Start(metricsAddr); err != nil {
 		common.Log.Fatalf("Failed to start metrics: %v", err)
 	}
@@ -159,7 +158,7 @@ func root(cmd *cobra.Command, args []string) {
 	handleError(resetIPSets(ips))
 	handleError(createBaseRules(ipt, ips))
 
-	npc := npc.New(nodeName, ipt, ips)
+	npc := npc.New(nodeName, legacy, ipt, ips)
 
 	nsController := makeController(client.Core().RESTClient(), "namespaces", &coreapi.Namespace{},
 		cache.ResourceEventHandlerFuncs{
@@ -201,25 +200,30 @@ func root(cmd *cobra.Command, args []string) {
 				handleError(npc.UpdatePod(old.(*coreapi.Pod), new.(*coreapi.Pod)))
 			}})
 
-	npController := makeController(client.Extensions().RESTClient(), "networkpolicies", &extnapi.NetworkPolicy{},
-		cache.ResourceEventHandlerFuncs{
-			AddFunc: func(obj interface{}) {
-				handleError(npc.AddNetworkPolicy(obj.(*extnapi.NetworkPolicy)))
-			},
-			DeleteFunc: func(obj interface{}) {
-				switch obj := obj.(type) {
-				case *extnapi.NetworkPolicy:
-					handleError(npc.DeleteNetworkPolicy(obj))
-				case cache.DeletedFinalStateUnknown:
-					// We know this object has gone away, but its final state is no longer
-					// available from the API server. Instead we use the last copy of it
-					// that we have, which is good enough for our cleanup.
-					handleError(npc.DeleteNetworkPolicy(obj.Obj.(*extnapi.NetworkPolicy)))
-				}
-			},
-			UpdateFunc: func(old, new interface{}) {
-				handleError(npc.UpdateNetworkPolicy(old.(*extnapi.NetworkPolicy), new.(*extnapi.NetworkPolicy)))
-			}})
+	npHandlers := cache.ResourceEventHandlerFuncs{
+		AddFunc: func(obj interface{}) {
+			handleError(npc.AddNetworkPolicy(obj))
+		},
+		DeleteFunc: func(obj interface{}) {
+			switch obj := obj.(type) {
+			case cache.DeletedFinalStateUnknown:
+				// We know this object has gone away, but its final state is no longer
+				// available from the API server. Instead we use the last copy of it
+				// that we have, which is good enough for our cleanup.
+				handleError(npc.DeleteNetworkPolicy(obj.Obj))
+			default:
+				handleError(npc.DeleteNetworkPolicy(obj))
+			}
+		},
+		UpdateFunc: func(old, new interface{}) {
+			handleError(npc.UpdateNetworkPolicy(old, new))
+		},
+	}
+	if legacy {
+		npController = makeController(client.Extensions().RESTClient(), "networkpolicies", &extnapi.NetworkPolicy{}, npHandlers)
+	} else {
+		npController = makeController(client.NetworkingV1().RESTClient(), "networkpolicies", &networkingv1.NetworkPolicy{}, npHandlers)
+	}
 
 	go nsController.Run(wait.NeverStop)
 	go podController.Run(wait.NeverStop)
@@ -240,6 +244,7 @@ func main() {
 	rootCmd.PersistentFlags().StringVar(&logLevel, "log-level", "debug", "logging level (debug, info, warning, error)")
 	rootCmd.PersistentFlags().BoolVar(&allowMcast, "allow-mcast", true, "allow all multicast traffic")
 	rootCmd.PersistentFlags().StringVar(&nodeName, "node-name", "", "only generate rules that apply to this node")
+	rootCmd.PersistentFlags().BoolVar(&legacy, "use-legacy-netpol", false, "use legacy network policies (pre k8s 1.7 vsn)")
 
 	handleError(rootCmd.Execute())
 }

prog/weavedb/Dockerfile

@@ -1,6 +1,7 @@
 # This is a nearly-empty image that we use to create a data-only container for persistence
 FROM scratch
 LABEL maintainer "Weaveworks Inc <help@weave.works>"
+LABEL works.weave.role=system
 ENTRYPOINT ["data-only"]
 # Work round Docker refusing to save an empty image
 COPY Dockerfile /

prog/weaveexec/Dockerfile.template

@@ -1,34 +1,9 @@
-FROM ALPINE_BASEIMAGE
+FROM DOCKERHUB_USER/weaveARCH_EXT
 
-# If we're building for another architecture than amd64, the CROSS_BUILD_ placeholder is removed so e.g. CROSS_BUILD_COPY turns into COPY
-# If we're building normally, for amd64, CROSS_BUILD lines are removed
-CROSS_BUILD_COPY qemu-QEMUARCH-static /usr/bin/
-
-LABEL maintainer "Weaveworks Inc <help@weave.works>"
-LABEL works.weave.role=system \
-      org.label-schema.name="Weave Net" \
-      org.label-schema.description="Weave Net creates a virtual network that connects Docker containers across multiple hosts and enables their automatic discovery" \
-      org.label-schema.url="https://weave.works" \
-      org.label-schema.vcs-url="https://github.com/weaveworks/weave" \
-      org.label-schema.vendor="Weaveworks"
-
-WORKDIR /home/weave
 ENTRYPOINT ["/home/weave/sigproxy", "/home/weave/weave"]
 
-RUN apk add --update \
-    curl \
-    ethtool \
-    iptables \
-    ipset \
-    iproute2 \
-    util-linux \
-    conntrack-tools \
-    bind-tools \
-  && rm -rf /var/cache/apk/*
-
-ADD ./weave ./sigproxy ./weaveproxy ./symlink /home/weave/
-ADD ./weaveutil /usr/bin/
+ADD ./sigproxy ./symlink /home/weave/
 ADD ./weavewait /w/w
 ADD ./weavewait_noop /w-noop/w
 ADD ./weavewait_nomcast /w-nomcast/w
-ADD ./docker /usr/bin/
+WORKDIR /home/weave

prog/weaveproxy/main.go

@@ -1,84 +0,0 @@
-package main
-
-import (
-	"fmt"
-	"os"
-	"strings"
-
-	"github.com/weaveworks/common/mflag"
-	"github.com/weaveworks/common/mflagext"
-	"github.com/weaveworks/common/signals"
-	"github.com/weaveworks/weave/common"
-	"github.com/weaveworks/weave/proxy"
-)
-
-var version = "unreleased"
-
-var Log = common.Log
-
-func getenv(key, defaultVal string) string {
-	if v := os.Getenv(key); v != "" {
-		return v
-	}
-	return defaultVal
-}
-
-func main() {
-	var (
-		justVersion bool
-		logLevel    = "info"
-		c           proxy.Config
-		withDNS     bool
-	)
-
-	c.Version = version
-
-	mflag.BoolVar(&justVersion, []string{"#version", "-version"}, false, "print version and exit")
-	mflag.StringVar(&logLevel, []string{"-log-level"}, "info", "logging level (debug, info, warning, error)")
-	mflagext.ListVar(&c.ListenAddrs, []string{"H"}, nil, "addresses on which to listen")
-	mflag.StringVar(&c.HostnameFromLabel, []string{"-hostname-from-label"}, "", "Key of container label from which to obtain the container's hostname")
-	mflag.StringVar(&c.HostnameMatch, []string{"-hostname-match"}, "(.*)", "Regexp pattern to apply on container names (e.g. '^aws-[0-9]+-(.*)$')")
-	mflag.StringVar(&c.HostnameReplacement, []string{"-hostname-replacement"}, "$1", "Expression to generate hostnames based on matches from --hostname-match (e.g. 'my-app-$1')")
-	mflag.BoolVar(&c.RewriteInspect, []string{"-rewrite-inspect"}, false, "Rewrite 'inspect' calls to return the weave network settings (if attached)")
-	mflag.BoolVar(&c.NoDefaultIPAM, []string{"#-no-default-ipam", "-no-default-ipalloc"}, false, "do not automatically allocate addresses for containers without a WEAVE_CIDR")
-	mflag.BoolVar(&c.NoRewriteHosts, []string{"-no-rewrite-hosts"}, false, "do not automatically rewrite /etc/hosts. Use if you need the docker IP to remain in /etc/hosts")
-	mflag.StringVar(&c.TLSConfig.CACert, []string{"#tlscacert", "-tlscacert"}, "", "Trust certs signed only by this CA")
-	mflag.StringVar(&c.TLSConfig.Cert, []string{"#tlscert", "-tlscert"}, "", "Path to TLS certificate file")
-	mflag.BoolVar(&c.TLSConfig.Enabled, []string{"#tls", "-tls"}, false, "Use TLS; implied by --tlsverify")
-	mflag.StringVar(&c.TLSConfig.Key, []string{"#tlskey", "-tlskey"}, "", "Path to TLS key file")
-	mflag.BoolVar(&c.TLSConfig.Verify, []string{"#tlsverify", "-tlsverify"}, false, "Use TLS and verify the remote")
-	mflag.BoolVar(&withDNS, []string{"#-with-dns", "#w"}, false, "option removed")
-	mflag.BoolVar(&c.WithoutDNS, []string{"-without-dns"}, false, "instruct created containers to never use weaveDNS as their nameserver")
-	mflag.BoolVar(&c.NoMulticastRoute, []string{"-no-multicast-route"}, false, "do not add a multicast route via the weave interface when attaching containers")
-	mflag.Parse()
-
-	if justVersion {
-		fmt.Printf("weave proxy  %s\n", version)
-		os.Exit(0)
-	}
-
-	common.SetLogLevel(logLevel)
-
-	Log.Infoln("weave proxy", version)
-	Log.Infoln("Command line arguments:", strings.Join(os.Args[1:], " "))
-
-	if withDNS {
-		Log.Warning("--with-dns option has been removed; DNS is on by default")
-	}
-
-	c.Image = getenv("EXEC_IMAGE", "weaveworks/weaveexec")
-	c.DockerBridge = getenv("DOCKER_BRIDGE", "docker0")
-	c.DockerHost = getenv("DOCKER_HOST", "unix:///var/run/docker.sock")
-
-	p, err := proxy.NewProxy(c)
-	if err != nil {
-		Log.Fatalf("Could not start proxy: %s", err)
-	}
-	defer p.Stop()
-
-	listeners := p.Listen()
-	p.AttachExistingContainers()
-	go p.Serve(listeners)
-	go p.ListenAndServeStatus("/home/weave/status.sock")
-	signals.SignalHandlerLoop(common.Log)
-}

prog/weaveproxy/main_test.go

@@ -1,12 +0,0 @@
-package main
-
-import (
-	"testing"
-
-	weavetest "github.com/weaveworks/weave/testing"
-)
-
-func TestMain(t *testing.T) {
-	weavetest.TrimTestArgs()
-	main()
-}

prog/weaver/Dockerfile.template

@@ -1,6 +1,31 @@
-FROM DOCKERHUB_USER/weaveexecARCH_EXT
-WORKDIR /home/weave
-ADD ./weaver /home/weave/
+FROM ALPINE_BASEIMAGE
+
+# If we're building for another architecture than amd64, the CROSS_BUILD_ placeholder is removed so e.g. CROSS_BUILD_COPY turns into COPY
+# If we're building normally, for amd64, CROSS_BUILD lines are removed
+CROSS_BUILD_COPY qemu-QEMUARCH-static /usr/bin/
+
+LABEL maintainer "Weaveworks Inc <help@weave.works>"
+LABEL works.weave.role=system \
+      org.label-schema.name="Weave Net" \
+      org.label-schema.description="Weave Net creates a virtual network that connects Docker containers across multiple hosts and enables their automatic discovery" \
+      org.label-schema.url="https://weave.works" \
+      org.label-schema.vcs-url="https://github.com/weaveworks/weave" \
+      org.label-schema.vendor="Weaveworks"
+
+RUN apk add --update \
+    curl \
+    ethtool \
+    iptables \
+    ipset \
+    iproute2 \
+    util-linux \
+    conntrack-tools \
+    bind-tools \
+  && rm -rf /var/cache/apk/*
+
+ADD ./weave ./weaver /home/weave/
+ADD ./weaveutil /usr/bin/
 ADD weavedata.db /weavedb/
 COPY ca-certificates.crt /etc/ssl/certs/
 ENTRYPOINT ["/home/weave/weaver"]
+WORKDIR /home/weave

prog/weaver/discovery.go

@@ -0,0 +1,104 @@
+package main
+
+import (
+	"bytes"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"io/ioutil"
+	"net/http"
+	"time"
+
+	"github.com/gorilla/mux"
+
+	"github.com/weaveworks/weave/ipam"
+)
+
+// TODO: move these definitions somewhere more shareable
+type PeerUpdateRequest struct {
+	Name      string   `json:"peername"`
+	Nickname  string   `json:"nickname"`  // optional
+	Addresses []string `json:"addresses"` // can be empty
+}
+
+type PeerUpdateResponse struct {
+	Addresses []string `json:"addresses"`
+	PeerCount int      `json:"peercount"`
+}
+
+var httpClient = &http.Client{Timeout: 30 * time.Second}
+
+func do(verb string, discoveryEndpoint, token string, request interface{}, response interface{}) error {
+	body := new(bytes.Buffer)
+	err := json.NewEncoder(body).Encode(request)
+	if err != nil {
+		return err
+	}
+	req, err := http.NewRequest(verb, discoveryEndpoint+"/peer", body)
+	if err != nil {
+		return err
+	}
+	req.Header.Set("Authorization", fmt.Sprintf("Scope-Probe token=%s", token))
+	req.Header.Set("X-Weave-Net-Version", version)
+	req.Header.Set("Content-Type", "application/json")
+	Log.Printf("Calling peer discovery %s with %s", req.URL, request)
+	resp, err := httpClient.Do(req)
+	if err != nil {
+		return err
+	}
+	defer resp.Body.Close()
+	if resp.StatusCode != http.StatusOK {
+		rbody, _ := ioutil.ReadAll(resp.Body)
+		return errors.New(resp.Status + ": " + string(rbody))
+	}
+	if response == nil {
+		return nil
+	}
+	err = json.NewDecoder(resp.Body).Decode(response)
+	Log.Printf("peer discovery result: (%v) %v", err, response)
+	return err
+}
+
+func peerDiscoveryUpdate(discoveryEndpoint, token, peername, nickname string, addresses []string) ([]string, int, error) {
+	request := PeerUpdateRequest{
+		Name:      peername,
+		Nickname:  nickname,
+		Addresses: addresses,
+	}
+	var updateResponse PeerUpdateResponse
+	err := do("POST", discoveryEndpoint, token, request, &updateResponse)
+	return updateResponse.Addresses, updateResponse.PeerCount, err
+}
+
+func peerDiscoveryDelete(discoveryEndpoint, token, peername string) error {
+	request := PeerUpdateRequest{Name: peername}
+	return do("DELETE", discoveryEndpoint, token, request, nil)
+}
+
+func HandleHTTPPeer(router *mux.Router, alloc *ipam.Allocator, discoveryEndpoint, token, peername string) {
+	router.Methods("DELETE").Path("/peer").HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if discoveryEndpoint != "" && token != "" {
+			if err := peerDiscoveryDelete(discoveryEndpoint, token, peername); err != nil {
+				Log.Errorf("Error while deleting self from peer discovery: %s", err)
+			}
+		}
+		if alloc != nil {
+			alloc.Shutdown()
+		}
+		w.WriteHeader(204)
+	})
+
+	router.Methods("DELETE").Path("/peer/{id}").HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		ident := mux.Vars(r)["id"]
+		if discoveryEndpoint != "" && token != "" {
+			// TODO: deal with this being either a peername or a nickname
+			if err := peerDiscoveryDelete(discoveryEndpoint, token, ident); err != nil {
+				Log.Errorf("Error while deleting self from peer discovery: %s", err)
+			}
+		}
+		if alloc != nil {
+			transferred := alloc.AdminTakeoverRanges(ident)
+			fmt.Fprintf(w, "%d IPs taken over from %s\n", transferred, ident)
+		}
+	})
+}

prog/weaver/http.go

@@ -13,9 +13,12 @@ import (
 	"github.com/weaveworks/mesh"
 
 	"github.com/weaveworks/go-checkpoint"
+	"github.com/weaveworks/weave/common"
 	"github.com/weaveworks/weave/ipam"
 	"github.com/weaveworks/weave/nameserver"
 	"github.com/weaveworks/weave/net/address"
+	"github.com/weaveworks/weave/plugin"
+	"github.com/weaveworks/weave/proxy"
 	weave "github.com/weaveworks/weave/router"
 )
 
@@ -31,23 +34,7 @@ var rootTemplate = template.New("root").Funcs(map[string]interface{}{
 	},
 	"printIPAMRanges": func(router weave.NetworkRouterStatus, status ipam.Status) string {
 		var buffer bytes.Buffer
-
-		type stats struct {
-			ips       uint32
-			nickname  string
-			reachable bool
-		}
-
-		peerStats := make(map[string]*stats)
-
-		for _, entry := range status.Entries {
-			s, found := peerStats[entry.Peer]
-			if !found {
-				s = &stats{nickname: entry.Nickname, reachable: entry.IsKnownPeer}
-				peerStats[entry.Peer] = s
-			}
-			s.ips += entry.Size
-		}
+		peerStats := summariseIpamStats(&status)
 
 		printOwned := func(name string, nickName string, info string, ips uint32) {
 			percentageRanges := float32(ips) * 100.0 / float32(status.RangeNumIPs)
@@ -113,6 +100,25 @@ var rootTemplate = template.New("root").Funcs(map[string]interface{}{
 	"trimSuffix": strings.TrimSuffix,
 })
 
+type ipamStats struct {
+	ips       uint32
+	nickname  string
+	reachable bool
+}
+
+func summariseIpamStats(status *ipam.Status) map[string]*ipamStats {
+	peerStats := make(map[string]*ipamStats)
+	for _, entry := range status.Entries {
+		s, found := peerStats[entry.Peer]
+		if !found {
+			s = &ipamStats{nickname: entry.Nickname, reachable: entry.IsKnownPeer}
+			peerStats[entry.Peer] = s
+		}
+		s.ips += entry.Size
+	}
+	return peerStats
+}
+
 func countDNSEntries(entries []nameserver.EntryStatus) int {
 	count := 0
 	for _, entry := range entries {
@@ -178,7 +184,7 @@ var statusTemplate = defTemplate("status", `\
 {{if allIPAMOwnersUnreachable .IPAM}}\
          Status: all IP ranges owned by unreachable peers - use 'rmpeer' if they are dead
 {{else if len .IPAM.PendingAllocates}}\
-         Status: waiting for IP range grant from peers
+         Status: waiting for IP(s) to become available
 {{else}}\
          Status: ready
 {{end}}\
@@ -202,6 +208,20 @@ var statusTemplate = defTemplate("status", `\
             TTL: {{.DNS.TTL}}
         Entries: {{countDNSEntries .DNS.Entries}}
 {{end}}\
+{{if .Proxy}}\
+
+        Service: proxy
+        Address: {{printList .Proxy.Addresses}}
+{{end}}\
+{{if .Plugin}}\
+
+{{if .Plugin.Config.EnableV2}}\
+        Service: plugin (v2)
+{{else}}\
+        Service: plugin (legacy)
+     DriverName: {{.Plugin.DriverName}}
+{{end}}\
+{{end}}\
 `)
 
 var targetsTemplate = defTemplate("targetsTemplate", `\
@@ -273,22 +293,29 @@ func (v *VersionCheck) String() string {
 }
 
 type WeaveStatus struct {
+	Ready        bool
 	Version      string
 	VersionCheck *VersionCheck              `json:"VersionCheck,omitempty"`
 	Router       *weave.NetworkRouterStatus `json:"Router,omitempty"`
 	IPAM         *ipam.Status               `json:"IPAM,omitempty"`
 	DNS          *nameserver.Status         `json:"DNS,omitempty"`
+	Proxy        *proxy.Status              `json:"Proxy,omitempty"`
+	Plugin       *plugin.Status             `json:"Plugin,omitempty"`
 }
 
 // Read-only functions, suitable for exposing on an unprotected socket
-func HandleHTTP(muxRouter *mux.Router, version string, router *weave.NetworkRouter, allocator *ipam.Allocator, defaultSubnet address.CIDR, ns *nameserver.Nameserver, dnsserver *nameserver.DNSServer) {
+func HandleHTTP(muxRouter *mux.Router, version string, router *weave.NetworkRouter, allocator *ipam.Allocator, defaultSubnet address.CIDR, ns *nameserver.Nameserver, dnsserver *nameserver.DNSServer, prxy *proxy.Proxy, plugin *plugin.Plugin, waitReady *common.WaitGroup) {
 	status := func() WeaveStatus {
 		return WeaveStatus{
+			waitReady.IsDone(),
 			version,
 			versionCheck(),
 			weave.NewNetworkRouterStatus(router),
 			ipam.NewStatus(allocator, defaultSubnet),
-			nameserver.NewStatus(ns, dnsserver)}
+			nameserver.NewStatus(ns, dnsserver),
+			proxy.NewStatus(prxy),
+			plugin.NewStatus(),
+		}
 	}
 	muxRouter.Methods("GET").Path("/report").Headers("Accept", "application/json").HandlerFunc(
 		func(w http.ResponseWriter, r *http.Request) {
@@ -324,7 +351,11 @@ func HandleHTTP(muxRouter *mux.Router, version string, router *weave.NetworkRout
 	defHandler := func(path string, template *template.Template) {
 		muxRouter.Methods("GET").Path(path).HandlerFunc(
 			func(w http.ResponseWriter, r *http.Request) {
-				if err := template.Execute(w, status()); err != nil {
+				s := status()
+				if !s.Ready {
+					w.WriteHeader(http.StatusServiceUnavailable)
+				}
+				if err := template.Execute(w, s); err != nil {
 					http.Error(w, "error during template execution", http.StatusInternalServerError)
 					Log.Error(err)
 				}

prog/weaver/main.go

@@ -14,6 +14,7 @@ import (
 	"github.com/gorilla/mux"
 	"github.com/pkg/profile"
 	"github.com/weaveworks/common/mflag"
+	"github.com/weaveworks/common/mflagext"
 	"github.com/weaveworks/common/signals"
 	"github.com/weaveworks/mesh"
 
@@ -25,6 +26,8 @@ import (
 	"github.com/weaveworks/weave/nameserver"
 	weavenet "github.com/weaveworks/weave/net"
 	"github.com/weaveworks/weave/net/address"
+	"github.com/weaveworks/weave/plugin"
+	weaveproxy "github.com/weaveworks/weave/proxy"
 	weave "github.com/weaveworks/weave/router"
 )
 
@@ -53,7 +56,7 @@ type dnsConfig struct {
 func (c *ipamConfig) Enabled() bool {
 	var (
 		hasPeerCount = c.PeerCount > 0
-		hasMode      = len(c.Mode) > 0
+		hasMode      = c.HasMode()
 		hasRange     = c.IPRangeCIDR != ""
 		hasSubnet    = c.IPSubnetCIDR != ""
 	)
@@ -63,9 +66,7 @@ func (c *ipamConfig) Enabled() bool {
 	case !hasRange && hasSubnet:
 		Log.Fatal("--ipalloc-default-subnet specified without --ipalloc-range.")
 	case !hasRange:
-		Log.Fatal("--ipalloc-init or --init-peer-count specified without --ipalloc-range.")
-	case hasMode && hasPeerCount:
-		Log.Fatal("At most one of --ipalloc-init or --init-peer-count may be specified.")
+		Log.Fatal("--ipalloc-init specified without --ipalloc-range.")
 	}
 	if hasMode {
 		if err := c.parseMode(); err != nil {
@@ -75,6 +76,10 @@ func (c *ipamConfig) Enabled() bool {
 	return true
 }
 
+func (c ipamConfig) HasMode() bool {
+	return len(c.Mode) > 0
+}
+
 func (c *ipamConfig) parseMode() error {
 	modeAndParam := strings.SplitN(c.Mode, "=", 2)
 
@@ -108,6 +113,13 @@ func (c *ipamConfig) parseMode() error {
 	return nil
 }
 
+func getenvOrDefault(key, defaultVal string) string {
+	if v := os.Getenv(key); v != "" {
+		return v
+	}
+	return defaultVal
+}
+
 func main() {
 	procs := runtime.NumCPU()
 	// packet sniffing can block an OS thread, so we need one thread
@@ -119,16 +131,17 @@ func main() {
 
 	var (
 		justVersion        bool
+		justPeerName       bool
 		config             mesh.Config
+		bridgeConfig       weavenet.BridgeConfig
 		networkConfig      weave.NetworkConfig
 		protocolMinVersion int
 		resume             bool
-		ifaceName          string
 		routerName         string
 		nickName           string
 		password           string
 		pktdebug           bool
-		logLevel           string
+		logLevel           = "info"
 		prof               string
 		bufSzMB            int
 		noDiscovery        bool
@@ -139,40 +152,42 @@ func main() {
 		peers              []string
 		noDNS              bool
 		dnsConfig          dnsConfig
-		datapathName       string
 		trustedSubnetStr   string
 		dbPrefix           string
-		isAWSVPC           bool
-
-		defaultDockerHost = "unix:///var/run/docker.sock"
+		hostRoot           string
+		procPath           string
+		discoveryEndpoint  string
+		token              string
+		advertiseAddress   string
+		pluginConfig       plugin.Config
+		defaultDockerHost  = getenvOrDefault("DOCKER_HOST", "unix:///var/run/docker.sock")
 	)
 
-	if val := os.Getenv("DOCKER_HOST"); val != "" {
-		defaultDockerHost = val
-	}
-
-	mflag.BoolVar(&justVersion, []string{"#version", "-version"}, false, "print version and exit")
+	mflag.BoolVar(&justVersion, []string{"-version"}, false, "print version and exit")
+	mflag.BoolVar(&justPeerName, []string{"-print-peer-name"}, false, "print peer name and exit")
 	mflag.StringVar(&config.Host, []string{"-host"}, "", "router host")
-	mflag.IntVar(&config.Port, []string{"#port", "-port"}, mesh.Port, "router port")
+	mflag.IntVar(&config.Port, []string{"-port"}, mesh.Port, "router port")
 	mflag.IntVar(&protocolMinVersion, []string{"-min-protocol-version"}, mesh.ProtocolMinVersion, "minimum weave protocol version")
 	mflag.BoolVar(&resume, []string{"-resume"}, false, "resume connections to previous peers")
-	mflag.StringVar(&ifaceName, []string{"#iface", "-iface"}, "", "name of interface to capture/inject from (disabled if blank)")
-	mflag.StringVar(&routerName, []string{"#name", "-name"}, "", "name of router (defaults to MAC of interface)")
-	mflag.StringVar(&nickName, []string{"#nickname", "-nickname"}, "", "nickname of peer (defaults to hostname)")
-	mflag.StringVar(&password, []string{"#password", "-password"}, "", "network password")
+	mflag.StringVar(&bridgeConfig.WeaveBridgeName, []string{"-weave-bridge"}, "weave", "name of weave bridge")
+	mflag.StringVar(&bridgeConfig.DockerBridgeName, []string{"-docker-bridge"}, "", "name of Docker bridge")
+	mflag.BoolVar(&bridgeConfig.NPC, []string{"-expect-npc"}, false, "set up iptables rules for npc")
+	mflag.StringVar(&routerName, []string{"-name"}, "", "name of router (defaults to MAC of interface)")
+	mflag.StringVar(&nickName, []string{"-nickname"}, "", "nickname of peer (defaults to hostname)")
+	mflag.StringVar(&password, []string{"-password"}, "", "network password")
 	mflag.StringVar(&logLevel, []string{"-log-level"}, "info", "logging level (debug, info, warning, error)")
-	mflag.BoolVar(&pktdebug, []string{"#pktdebug", "#-pktdebug", "-pkt-debug"}, false, "enable per-packet debug logging")
-	mflag.StringVar(&prof, []string{"#profile", "-profile"}, "", "enable profiling and write profiles to given path")
-	mflag.IntVar(&config.ConnLimit, []string{"#connlimit", "#-connlimit", "-conn-limit"}, 30, "connection limit (0 for unlimited)")
-	mflag.BoolVar(&noDiscovery, []string{"#nodiscovery", "#-nodiscovery", "-no-discovery"}, false, "disable peer discovery")
-	mflag.IntVar(&bufSzMB, []string{"#bufsz", "-bufsz"}, 8, "capture buffer size in MB")
-	mflag.StringVar(&httpAddr, []string{"#httpaddr", "#-httpaddr", "-http-addr"}, "", "address to bind HTTP interface to (disabled if blank, absolute path indicates unix domain socket)")
+	mflag.BoolVar(&pktdebug, []string{"-pkt-debug"}, false, "enable per-packet debug logging")
+	mflag.StringVar(&prof, []string{"-profile"}, "", "enable profiling and write profiles to given path")
+	mflag.IntVar(&config.ConnLimit, []string{"-conn-limit"}, 30, "connection limit (0 for unlimited)")
+	mflag.BoolVar(&noDiscovery, []string{"-no-discovery"}, false, "disable peer discovery")
+	mflag.IntVar(&bufSzMB, []string{"-bufsz"}, 8, "capture buffer size in MB")
+	mflag.IntVar(&bridgeConfig.MTU, []string{"-mtu"}, 0, "MTU size")
+	mflag.StringVar(&httpAddr, []string{"-http-addr"}, "", "address to bind HTTP interface to (disabled if blank, absolute path indicates unix domain socket)")
 	mflag.StringVar(&statusAddr, []string{"-status-addr"}, "", "address to bind status+metrics interface to (disabled if blank, absolute path indicates unix domain socket)")
 	mflag.StringVar(&ipamConfig.Mode, []string{"-ipalloc-init"}, "", "allocator initialisation strategy (consensus, seed or observer)")
-	mflag.StringVar(&ipamConfig.IPRangeCIDR, []string{"#iprange", "#-iprange", "-ipalloc-range"}, "", "IP address range reserved for automatic allocation, in CIDR notation")
-	mflag.StringVar(&ipamConfig.IPSubnetCIDR, []string{"#ipsubnet", "#-ipsubnet", "-ipalloc-default-subnet"}, "", "subnet to allocate within by default, in CIDR notation")
-	mflag.IntVar(&ipamConfig.PeerCount, []string{"#initpeercount", "#-initpeercount", "-init-peer-count"}, 0, "number of peers in network (for IP address allocation)")
-	mflag.StringVar(&dockerAPI, []string{"#api", "#-api", "-docker-api"}, defaultDockerHost, "Docker API endpoint")
+	mflag.StringVar(&ipamConfig.IPRangeCIDR, []string{"-ipalloc-range"}, "", "IP address range reserved for automatic allocation, in CIDR notation")
+	mflag.StringVar(&ipamConfig.IPSubnetCIDR, []string{"-ipalloc-default-subnet"}, "", "subnet to allocate within by default, in CIDR notation")
+	mflag.StringVar(&dockerAPI, []string{"-docker-api"}, defaultDockerHost, "Docker API endpoint")
 	mflag.BoolVar(&noDNS, []string{"-no-dns"}, false, "disable DNS server")
 	mflag.StringVar(&dnsConfig.Domain, []string{"-dns-domain"}, nameserver.DefaultDomain, "local domain to server requests for")
 	mflag.StringVar(&dnsConfig.ListenAddress, []string{"-dns-listen-address"}, nameserver.DefaultListenAddress, "address to listen on for DNS requests")
@@ -180,10 +195,24 @@ func main() {
 	mflag.DurationVar(&dnsConfig.ClientTimeout, []string{"-dns-fallback-timeout"}, nameserver.DefaultClientTimeout, "timeout for fallback DNS requests")
 	mflag.StringVar(&dnsConfig.EffectiveListenAddress, []string{"-dns-effective-listen-address"}, "", "address DNS will actually be listening, after Docker port mapping")
 	mflag.StringVar(&dnsConfig.ResolvConf, []string{"-resolv-conf"}, "", "path to resolver configuration for fallback DNS lookups")
-	mflag.StringVar(&datapathName, []string{"-datapath"}, "", "ODP datapath name")
+	mflag.StringVar(&bridgeConfig.DatapathName, []string{"-datapath"}, "", "ODP datapath name")
+	mflag.BoolVar(&bridgeConfig.NoFastdp, []string{"-no-fastdp"}, false, "Disable Fast Datapath")
 	mflag.StringVar(&trustedSubnetStr, []string{"-trusted-subnets"}, "", "comma-separated list of trusted subnets in CIDR notation")
 	mflag.StringVar(&dbPrefix, []string{"-db-prefix"}, "/weavedb/weave", "pathname/prefix of filename to store data")
-	mflag.BoolVar(&isAWSVPC, []string{"-awsvpc"}, false, "use AWS VPC for routing")
+	mflag.StringVar(&procPath, []string{"-proc-path"}, "/proc", "path to reach host /proc filesystem")
+	mflag.BoolVar(&bridgeConfig.AWSVPC, []string{"-awsvpc"}, false, "use AWS VPC for routing")
+	mflag.StringVar(&hostRoot, []string{"-host-root"}, "", "path to reach host filesystem")
+	mflag.StringVar(&discoveryEndpoint, []string{"-peer-discovery-url"}, "https://cloud.weave.works/api/net", "url for peer discovery")
+	mflag.StringVar(&token, []string{"-token"}, "", "token for peer discovery")
+	mflag.StringVar(&advertiseAddress, []string{"-advertise-address"}, "", "address to advertise for peer discovery")
+
+	mflag.BoolVar(&pluginConfig.Enable, []string{"-plugin"}, false, "enable Docker plugin (v1)")
+	mflag.BoolVar(&pluginConfig.EnableV2, []string{"-plugin-v2"}, false, "enable Docker plugin (v2)")
+	mflag.BoolVar(&pluginConfig.EnableV2Multicast, []string{"-plugin-v2-multicast"}, false, "enable multicast for Docker plugin (v2)")
+	mflag.StringVar(&pluginConfig.Socket, []string{"-plugin-socket"}, "/run/docker/plugins/weave.sock", "plugin socket on which to listen")
+	mflag.StringVar(&pluginConfig.MeshSocket, []string{"-plugin-mesh-socket"}, "/run/docker/plugins/weavemesh.sock", "plugin socket on which to listen in mesh mode")
+
+	proxyConfig := newProxyConfig()
 
 	// crude way of detecting that we probably have been started in a
 	// container, with `weave launch` --> suppress misleading paths in
@@ -195,19 +224,30 @@ func main() {
 
 	mflag.Parse()
 
+	if justVersion {
+		fmt.Printf("weave %s\n", version)
+		os.Exit(0)
+	}
+	name := peerName(routerName, bridgeConfig.WeaveBridgeName, dbPrefix, hostRoot)
+	if justPeerName {
+		fmt.Printf("%s\n", name)
+		os.Exit(0)
+	}
+
+	proxyConfig.DockerHost = dockerAPI
+	if bridgeConfig.AWSVPC {
+		proxyConfig.NoMulticastRoute = true
+		proxyConfig.KeepTXOn = true
+	}
+
 	peers = mflag.Args()
 	if resume && len(peers) > 0 {
 		Log.Fatalf("You must not specify an initial peer list in conjunction with --resume")
 	}
 
 	common.SetLogLevel(logLevel)
-
-	if justVersion {
-		fmt.Printf("weave router %s\n", version)
-		os.Exit(0)
-	}
-
 	Log.Println("Command line options:", options())
+	Log.Infoln("weave ", version)
 
 	if prof != "" {
 		defer profile.Start(profile.CPUProfile, profile.ProfilePath(prof), profile.NoShutdownHook).Stop()
@@ -218,26 +258,57 @@ func main() {
 	}
 	config.ProtocolMinVersion = byte(protocolMinVersion)
 
+	var waitReady common.WaitGroup
+
+	var proxy *weaveproxy.Proxy
+	var err error
+	if proxyConfig.Enabled {
+		if noDNS {
+			proxyConfig.WithoutDNS = true
+		}
+		// Start Weave Proxy:
+		proxy, err = weaveproxy.NewProxy(*proxyConfig)
+		if err != nil {
+			Log.Fatalf("Could not start proxy: %s", err)
+		}
+		defer proxy.Stop()
+		listeners := proxy.Listen()
+		proxy.AttachExistingContainers()
+		go proxy.Serve(listeners, waitReady.Add())
+	}
+
 	if pktdebug {
 		networkConfig.PacketLogging = packetLogging{}
 	} else {
 		networkConfig.PacketLogging = nopPacketLogging{}
 	}
 
+	if bridgeConfig.DockerBridgeName != "" {
+		if setAddr, err := weavenet.EnforceAddrAssignType(bridgeConfig.DockerBridgeName); err != nil {
+			Log.Errorf("While checking address assignment type of %s: %s", bridgeConfig.DockerBridgeName, err)
+		} else if setAddr {
+			Log.Warningf("Setting %s MAC (mitigate https://github.com/docker/docker/issues/14908)", bridgeConfig.DockerBridgeName)
+		}
+	}
+
+	bridgeConfig.Mac = name.String()
+	bridgeConfig.Port = config.Port
+	bridgeType, err := weavenet.EnsureBridge(procPath, &bridgeConfig, Log)
+	checkFatal(err)
+	Log.Println("Bridge type is", bridgeType)
+
 	config.Password = determinePassword(password)
 
-	overlay, bridge := createOverlay(datapathName, ifaceName, isAWSVPC, config.Host, config.Port, bufSzMB, config.Password != nil)
-	networkConfig.Bridge = bridge
+	overlay, injectorConsumer := createOverlay(bridgeType, bridgeConfig, config.Host, config.Port, bufSzMB, config.Password != nil)
+	networkConfig.InjectorConsumer = injectorConsumer
 
-	if bridge != nil {
+	if injectorConsumer != nil {
 		if err := weavenet.DetectHairpin("vethwe-bridge", Log); err != nil {
 			Log.Errorf("Setting may cause connectivity issues : %s", err)
 			Log.Infof("Hairpin mode may have been enabled by other software on this machine")
 		}
 	}
 
-	name := peerName(routerName)
-
 	if nickName == "" {
 		var err error
 		nickName, err = os.Hostname()
@@ -247,19 +318,39 @@ func main() {
 	config.TrustedSubnets = parseTrustedSubnets(trustedSubnetStr)
 	config.PeerDiscovery = !noDiscovery
 
-	if isAWSVPC && len(config.Password) > 0 {
+	if bridgeConfig.AWSVPC && len(config.Password) > 0 {
 		Log.Fatalf("--awsvpc mode is not compatible with the --password option")
 	}
+	if bridgeConfig.AWSVPC && !ipamConfig.Enabled() {
+		Log.Fatalf("--awsvpc mode requires IPAM enabled")
+	}
 
-	db, err := db.NewBoltDB(dbPrefix + db.FileName)
+	db, err := db.NewBoltDB(dbPrefix)
 	checkFatal(err)
 	defer db.Close()
 
-	router, err := weave.NewNetworkRouter(config, networkConfig, name, nickName, overlay, db)
+	router, err := weave.NewNetworkRouter(config, networkConfig, bridgeConfig, name, nickName, overlay, db)
 	checkFatal(err)
 	Log.Println("Our name is", router.Ourself)
 
-	if peers, err = router.InitialPeers(resume, peers); err != nil {
+	if token != "" {
+		var addresses []string
+		if advertiseAddress == "" {
+			localAddrs, err := weavenet.LocalAddresses()
+			checkFatal(err)
+			for _, addr := range localAddrs {
+				addresses = append(addresses, addr.IP.String())
+			}
+		} else {
+			addresses = strings.Split(advertiseAddress, ",")
+		}
+		discoveredPeers, count, err := peerDiscoveryUpdate(discoveryEndpoint, token, name.String(), nickName, addresses)
+		checkFatal(err)
+		if !ipamConfig.HasMode() {
+			ipamConfig.PeerCount = len(peers) + count
+		}
+		peers = append(peers, discoveredPeers...)
+	} else if peers, err = router.InitialPeers(resume, peers); err != nil {
 		Log.Fatal("Unable to get initial peer set: ", err)
 	}
 
@@ -296,24 +387,26 @@ func main() {
 	)
 	if ipamConfig.Enabled() {
 		var t tracker.LocalRangeTracker
-		if isAWSVPC {
+		if bridgeConfig.AWSVPC {
 			Log.Infoln("Creating AWSVPC LocalRangeTracker")
-			t, err = tracker.NewAWSVPCTracker()
+			t, err = tracker.NewAWSVPCTracker(bridgeConfig.WeaveBridgeName)
 			if err != nil {
 				Log.Fatalf("Cannot create AWSVPC LocalRangeTracker: %s", err)
 			}
 			trackerName = "awsvpc"
 		}
-		var allContainerIDs []string
-		if dockerCli != nil {
-			allContainerIDs, err = dockerCli.AllContainerIDs()
-			checkFatal(err)
-		}
-		preClaims, err := findExistingAddresses(dockerCli, allContainerIDs, weavenet.WeaveBridgeName)
+
+		preClaims, err := findExistingAddresses(dockerCli, bridgeConfig.WeaveBridgeName)
 		checkFatal(err)
+
 		allocator, defaultSubnet = createAllocator(router, ipamConfig, preClaims, db, t, isKnownPeer)
 		observeContainers(allocator)
-		allocator.PruneOwned(allContainerIDs)
+
+		if dockerCli != nil {
+			allContainerIDs, err := dockerCli.RunningContainerIDs()
+			checkFatal(err)
+			allocator.PruneOwned(allContainerIDs)
+		}
 	}
 
 	var (
@@ -326,6 +419,9 @@ func main() {
 		ns.Start()
 		defer ns.Stop()
 		dnsserver.ActivateAndServe()
+		if dockerCli != nil {
+			populateDNS(ns, dockerCli, name, bridgeConfig.WeaveBridgeName)
+		}
 		defer dnsserver.Stop()
 	}
 
@@ -335,6 +431,10 @@ func main() {
 	}
 	checkFatal(router.CreateRestartSentinel())
 
+	pluginConfig.DNS = !noDNS
+	pluginConfig.DefaultSubnet = defaultSubnet.String()
+	plugin := plugin.NewPlugin(pluginConfig)
+
 	// The weave script always waits for a status call to succeed,
 	// so there is no point in doing "weave launch --http-addr ''".
 	// This is here to support stand-alone use of weaver.
@@ -347,8 +447,12 @@ func main() {
 			ns.HandleHTTP(muxRouter, dockerCli)
 		}
 		router.HandleHTTP(muxRouter)
-		HandleHTTP(muxRouter, version, router, allocator, defaultSubnet, ns, dnsserver)
+		HandleHTTP(muxRouter, version, router, allocator, defaultSubnet, ns, dnsserver, proxy, plugin, &waitReady)
+		HandleHTTPPeer(muxRouter, allocator, discoveryEndpoint, token, name.String())
 		muxRouter.Methods("GET").Path("/metrics").Handler(metricsHandler(router, allocator, ns, dnsserver))
+		if proxy != nil {
+			muxRouter.Methods("GET").Path("/proxyaddrs").HandlerFunc(proxy.StatusHTTP)
+		}
 		http.Handle("/", common.LoggingHTTPHandler(muxRouter))
 		Log.Println("Listening for HTTP control messages on", httpAddr)
 		go listenAndServeHTTP(httpAddr, nil)
@@ -356,7 +460,7 @@ func main() {
 
 	if statusAddr != "" {
 		muxRouter := mux.NewRouter()
-		HandleHTTP(muxRouter, version, router, allocator, defaultSubnet, ns, dnsserver)
+		HandleHTTP(muxRouter, version, router, allocator, defaultSubnet, ns, dnsserver, proxy, plugin, &waitReady)
 		muxRouter.Methods("GET").Path("/metrics").Handler(metricsHandler(router, allocator, ns, dnsserver))
 		statusMux := http.NewServeMux()
 		statusMux.Handle("/", muxRouter)
@@ -364,16 +468,37 @@ func main() {
 		go listenAndServeHTTP(statusAddr, statusMux)
 	}
 
+	if plugin != nil {
+		go plugin.Start(httpAddr, dockerCli, waitReady.Add())
+	}
+
+	if bridgeConfig.AWSVPC {
+		// Run this on its own goroutine because the allocator can block
+		// We remove the default route installed by the kernel,
+		// because awsvpc has installed it as well
+		go exposeForAWSVPC(allocator, defaultSubnet, bridgeConfig.WeaveBridgeName, waitReady.Add())
+	}
+
 	signals.SignalHandlerLoop(common.Log, router)
 }
 
+func exposeForAWSVPC(alloc *ipam.Allocator, subnet address.CIDR, bridgeName string, ready func()) {
+	addr, err := alloc.Allocate("weave:expose", subnet, false, func() bool { return false })
+	checkFatal(err)
+	cidr := address.MakeCIDR(subnet, addr)
+	err = weavenet.Expose(bridgeName, cidr.IPNet(), true, false)
+	checkFatal(err)
+	Log.Printf("Bridge %q exposed on address %v", bridgeName, cidr)
+	ready()
+}
+
 func options() map[string]string {
 	options := make(map[string]string)
 	mflag.Visit(func(f *mflag.Flag) {
 		value := f.Value.String()
 		name := canonicalName(f)
-		if name == "password" {
-			value = "<elided>"
+		if name == "password" || name == "token" {
+			value = "<redacted>"
 		}
 		options[name] = value
 	})
@@ -407,34 +532,55 @@ func (nopPacketLogging) LogPacket(string, weave.PacketKey) {
 func (nopPacketLogging) LogForwardPacket(string, weave.ForwardPacketKey) {
 }
 
-func createOverlay(datapathName string, ifaceName string, isAWSVPC bool, host string, port int, bufSzMB int, enableEncryption bool) (weave.NetworkOverlay, weave.Bridge) {
+func newProxyConfig() *weaveproxy.Config {
+	proxyConfig := weaveproxy.Config{
+		Image:        getenvOrDefault("EXEC_IMAGE", "weaveworks/weaveexec"),
+		DockerBridge: getenvOrDefault("DOCKER_BRIDGE", "docker0"),
+	}
+	mflag.BoolVar(&proxyConfig.Enabled, []string{"-proxy"}, false, "instruct Weave Net to start its Docker proxy")
+	mflagext.ListVar(&proxyConfig.ListenAddrs, []string{"H"}, nil, "addresses on which to listen for Docker proxy")
+	mflag.StringVar(&proxyConfig.HostnameFromLabel, []string{"-hostname-from-label"}, "", "Key of container label from which to obtain the container's hostname")
+	mflag.StringVar(&proxyConfig.HostnameMatch, []string{"-hostname-match"}, "(.*)", "Regexp pattern to apply on container names (e.g. '^aws-[0-9]+-(.*)$')")
+	mflag.StringVar(&proxyConfig.HostnameReplacement, []string{"-hostname-replacement"}, "$1", "Expression to generate hostnames based on matches from --hostname-match (e.g. 'my-app-$1')")
+	mflag.BoolVar(&proxyConfig.RewriteInspect, []string{"-rewrite-inspect"}, false, "Rewrite 'inspect' calls to return the weave network settings (if attached)")
+	mflag.BoolVar(&proxyConfig.NoDefaultIPAM, []string{"-no-default-ipalloc"}, false, "proxy: do not automatically allocate addresses for containers without a WEAVE_CIDR")
+	mflag.BoolVar(&proxyConfig.NoRewriteHosts, []string{"-no-rewrite-hosts"}, false, "proxy: do not automatically rewrite /etc/hosts. Use if you need the docker IP to remain in /etc/hosts")
+	mflag.StringVar(&proxyConfig.TLSConfig.CACert, []string{"-tlscacert"}, "", "Trust certs signed only by this CA")
+	mflag.StringVar(&proxyConfig.TLSConfig.Cert, []string{"-tlscert"}, "", "Path to TLS certificate file")
+	mflag.BoolVar(&proxyConfig.TLSConfig.Enabled, []string{"-tls"}, false, "Use TLS; implied by --tlsverify")
+	mflag.StringVar(&proxyConfig.TLSConfig.Key, []string{"-tlskey"}, "", "Path to TLS key file")
+	mflag.BoolVar(&proxyConfig.TLSConfig.Verify, []string{"-tlsverify"}, false, "Use TLS and verify the remote")
+	mflag.BoolVar(&proxyConfig.WithoutDNS, []string{"-without-dns"}, false, "proxy: instruct created containers to never use weaveDNS as their nameserver")
+	mflag.BoolVar(&proxyConfig.NoMulticastRoute, []string{"-no-multicast-route"}, false, "proxy: do not add a multicast route via the weave interface when attaching containers")
+	return &proxyConfig
+}
+
+func createOverlay(bridgeType weavenet.Bridge, config weavenet.BridgeConfig, host string, port int, bufSzMB int, enableEncryption bool) (weave.NetworkOverlay, weave.InjectorConsumer) {
 	overlay := weave.NewOverlaySwitch()
-	var bridge weave.Bridge
+	var injectorConsumer weave.InjectorConsumer
 	var ignoreSleeve bool
 
 	switch {
-	case isAWSVPC:
+	case config.AWSVPC:
 		vpc := weave.NewAWSVPC()
 		overlay.Add("awsvpc", vpc)
-		bridge = weave.NullBridge{}
+		injectorConsumer = weave.NullInjectorConsumer{}
 		// Currently, we do not support any overlay with AWSVPC
 		ignoreSleeve = true
-	case datapathName != "" && ifaceName != "":
-		Log.Fatal("At most one of --datapath and --iface must be specified.")
-	case datapathName != "":
-		iface, err := weavenet.EnsureInterface(datapathName)
+	case bridgeType == nil:
+		injectorConsumer = weave.NullInjectorConsumer{}
+	case bridgeType.IsFastdp():
+		iface, err := weavenet.EnsureInterface(config.DatapathName)
 		checkFatal(err)
 		fastdp, err := weave.NewFastDatapath(iface, port, enableEncryption)
 		checkFatal(err)
-		bridge = fastdp.Bridge()
+		injectorConsumer = fastdp.InjectorConsumer()
 		overlay.Add("fastdp", fastdp.Overlay())
-	case ifaceName != "":
-		iface, err := weavenet.EnsureInterface(ifaceName)
+	case !bridgeType.IsFastdp():
+		iface, err := weavenet.EnsureInterface(weavenet.PcapIfName)
 		checkFatal(err)
-		bridge, err = weave.NewPcap(iface, bufSzMB*1024*1024) // bufsz flag is in MB
+		injectorConsumer, err = weave.NewPcap(iface, bufSzMB*1024*1024) // bufsz flag is in MB
 		checkFatal(err)
-	default:
-		bridge = weave.NullBridge{}
 	}
 
 	if !ignoreSleeve {
@@ -443,7 +589,7 @@ func createOverlay(datapathName string, ifaceName string, isAWSVPC bool, host st
 		overlay.SetCompatOverlay(sleeve)
 	}
 
-	return overlay, bridge
+	return overlay, injectorConsumer
 }
 
 func createAllocator(router *weave.NetworkRouter, config ipamConfig, preClaims []ipam.PreClaim, db db.DB, track tracker.LocalRangeTracker, isKnownPeer func(mesh.PeerName) bool) (*ipam.Allocator, address.CIDR) {
@@ -536,13 +682,15 @@ func determinePassword(password string) []byte {
 	return []byte(password)
 }
 
-func peerName(routerName string) mesh.PeerName {
+func peerName(routerName, bridgeName, dbPrefix, hostRoot string) mesh.PeerName {
 	if routerName == "" {
-		iface, err := net.InterfaceByName(weavenet.WeaveBridgeName)
-		if err != nil {
-			Log.Fatalf("Unable to find bridge %q", weavenet.WeaveBridgeName)
+		iface, err := net.InterfaceByName(bridgeName)
+		if err == nil {
+			routerName = iface.HardwareAddr.String()
+		} else {
+			routerName, err = weavenet.GetSystemPeerName(dbPrefix, hostRoot)
+			checkFatal(err)
 		}
-		routerName = iface.HardwareAddr.String()
 	}
 	name, err := mesh.PeerNameFromUserInput(routerName)
 	checkFatal(err)

prog/weaver/main_test.go

@@ -7,6 +7,7 @@ import (
 )
 
 func TestMain(t *testing.T) {
-	weavetest.TrimTestArgs()
-	main()
+	if weavetest.TrimTestArgs() {
+		main()
+	}
 }

prog/weaver/metrics.go

@@ -82,6 +82,27 @@ var metrics = []metric{
 				ch <- intGauge(desc, metrics.Flows)
 			}
 		}},
+	{desc("weave_ipam_unreachable_count", "Number of unreachable peers."),
+		func(s WeaveStatus, desc *prometheus.Desc, ch chan<- prometheus.Metric) {
+			var count int
+			for _, entry := range summariseIpamStats(s.IPAM) {
+				if !entry.reachable {
+					count++
+				}
+			}
+			ch <- intGauge(desc, count)
+		}},
+	{desc("weave_ipam_unreachable_percentage", "Percentage of IP addresses owned  by unreachable peers."),
+		func(s WeaveStatus, desc *prometheus.Desc, ch chan<- prometheus.Metric) {
+			var totalUnreachable uint32
+			for _, entry := range summariseIpamStats(s.IPAM) {
+				if !entry.reachable {
+					totalUnreachable += entry.ips
+				}
+			}
+			percentage := float64(totalUnreachable) * 100.0 / float64(s.IPAM.RangeNumIPs)
+			ch <- prometheus.MustNewConstMetric(desc, prometheus.GaugeValue, percentage)
+		}},
 	{desc("weave_ipam_pending_allocates", "Number of pending allocates."),
 		func(s WeaveStatus, desc *prometheus.Desc, ch chan<- prometheus.Metric) {
 			if s.IPAM != nil {
@@ -118,10 +139,10 @@ func newMetrics(router *weave.NetworkRouter, allocator *ipam.Allocator, ns *name
 
 func (m *collector) Collect(ch chan<- prometheus.Metric) {
 
-	status := WeaveStatus{"", nil,
-		weave.NewNetworkRouterStatus(m.router),
-		ipam.NewStatus(m.allocator, address.CIDR{}),
-		nameserver.NewStatus(m.ns, m.dnsserver)}
+	status := WeaveStatus{
+		Router: weave.NewNetworkRouterStatus(m.router),
+		IPAM:   ipam.NewStatus(m.allocator, address.CIDR{}),
+		DNS:    nameserver.NewStatus(m.ns, m.dnsserver)}
 
 	for _, metric := range metrics {
 		metric.Collect(status, metric.Desc, ch)

prog/weaver/reclaim.go

@@ -2,13 +2,16 @@ package main
 
 import (
 	"net"
+	"strings"
 
 	docker "github.com/fsouza/go-dockerclient"
 
+	"github.com/weaveworks/mesh"
 	"github.com/weaveworks/weave/api"
 	"github.com/weaveworks/weave/common"
 	weavedocker "github.com/weaveworks/weave/common/docker"
 	"github.com/weaveworks/weave/ipam"
+	"github.com/weaveworks/weave/nameserver"
 	weavenet "github.com/weaveworks/weave/net"
 	"github.com/weaveworks/weave/net/address"
 )
@@ -20,7 +23,7 @@ func a(cidr *net.IPNet) address.CIDR {
 
 // Get all the existing Weave IPs at startup, so we can stop IPAM
 // giving out any as duplicates
-func findExistingAddresses(dockerCli *weavedocker.Client, containerIDs []string, bridgeName string) (addrs []ipam.PreClaim, err error) {
+func findExistingAddresses(dockerCli *weavedocker.Client, bridgeName string) (addrs []ipam.PreClaim, err error) {
 	Log.Infof("Checking for pre-existing addresses on %s bridge", bridgeName)
 	// First get the address for the bridge
 	bridgeNetDev, err := weavenet.GetBridgeNetDev(bridgeName)
@@ -50,6 +53,11 @@ func findExistingAddresses(dockerCli *weavedocker.Client, containerIDs []string,
 	// Now iterate over all containers to see if they have a network
 	// namespace with an attached interface
 	if dockerCli != nil {
+		containerIDs, err := dockerCli.RunningContainerIDs()
+		if err != nil {
+			return nil, err
+		}
+
 		for _, cid := range containerIDs {
 			container, err := dockerCli.InspectContainer(cid)
 			if err != nil {
@@ -82,3 +90,49 @@ func findExistingAddresses(dockerCli *weavedocker.Client, containerIDs []string,
 	}
 	return addrs, nil
 }
+
+func populateDNS(ns *nameserver.Nameserver, dockerCli *weavedocker.Client, ourName mesh.PeerName, bridgeName string) (addrs []ipam.PreClaim, err error) {
+	// Find all veths connected to the bridge
+	peerIDs, err := weavenet.ConnectedToBridgeVethPeerIds(bridgeName)
+	if err != nil {
+		return nil, err
+	}
+
+	// Now iterate over all containers to see if they have a network
+	// namespace with an attached interface
+	if dockerCli != nil {
+		containerIDs, err := dockerCli.RunningContainerIDs()
+		if err != nil {
+			return nil, err
+		}
+
+		for _, cid := range containerIDs {
+			container, err := dockerCli.InspectContainer(cid)
+			if err != nil {
+				if _, ok := err.(*docker.NoSuchContainer); ok {
+					continue
+				}
+				return nil, err
+			}
+			if container.State.Pid != 0 {
+				// Docker allows containers to have fqdns in two wauys:
+				// where the container explicitly has a domain name, and when its
+				// hostname contains the domain name, recognized by it having dots.
+				if container.Config.Domainname == "" && !strings.Contains(container.Config.Hostname, ".") {
+					continue
+				}
+				netDevs, err := weavenet.GetNetDevsByVethPeerIds(container.State.Pid, peerIDs)
+				if err != nil {
+					return nil, err
+				}
+				fqdn := container.Config.Hostname + "." + container.Config.Domainname
+				for _, netDev := range netDevs {
+					for _, cidr := range netDev.CIDRs {
+						ns.AddEntryFQDN(fqdn, cid, ourName, address.FromIP4(cidr.IP))
+					}
+				}
+			}
+		}
+	}
+	return addrs, nil
+}

prog/weaveutil/addrs.go

@@ -27,10 +27,25 @@ func containerAddrs(args []string) error {
 		return err
 	}
 
+	containerArgs := []string{}
+	for _, cid := range args[1:] {
+		if cid == "weave:allids" { // expand to all container IDs
+			all, err := client.ListContainers(docker.ListContainersOptions{})
+			if err != nil {
+				return err
+			}
+			for _, c := range all {
+				containerArgs = append(containerArgs, c.ID)
+			}
+		} else {
+			containerArgs = append(containerArgs, cid)
+		}
+	}
+
 	var containerIDs []string
 	containers := make(map[string]*docker.Container)
 
-	for _, cid := range args[1:] {
+	for _, cid := range containerArgs {
 		if cid == "weave:expose" {
 			netDev, err := weavenet.GetBridgeNetDev(bridgeName)
 			if err != nil {

prog/weaveutil/attach.go

@@ -3,13 +3,15 @@ package main
 import (
 	"fmt"
 	"net"
+	"os"
 	"strconv"
+	"strings"
 	"syscall"
 
-	docker "github.com/fsouza/go-dockerclient"
 	"github.com/vishvananda/netns"
 
 	weavenet "github.com/weaveworks/weave/net"
+	"github.com/weaveworks/weave/proxy"
 )
 
 func attach(args []string) error {
@@ -67,21 +69,6 @@ func attach(args []string) error {
 	return err
 }
 
-func containerPid(containerID string) (int, error) {
-	c, err := docker.NewVersionedClientFromEnv("1.18")
-	if err != nil {
-		return 0, fmt.Errorf("unable to connect to docker: %s", err)
-	}
-	container, err := c.InspectContainer(containerID)
-	if err != nil {
-		return 0, fmt.Errorf("unable to inspect container %s: %s", containerID, err)
-	}
-	if container.State.Pid == 0 {
-		return 0, fmt.Errorf("container %s not running", containerID)
-	}
-	return container.State.Pid, nil
-}
-
 func processExists(pid int) bool {
 	err := syscall.Kill(pid, syscall.Signal(0))
 	return err == nil || err == syscall.EPERM
@@ -114,3 +101,40 @@ func detach(args []string) error {
 	}
 	return weavenet.DetachContainer(weavenet.NSPathByPid(pid), args[0], weavenet.VethName, cidrs)
 }
+
+func rewriteEtcHosts(args []string) error {
+	if len(args) < 3 {
+		cmdUsage("rewrite-etc-hosts", "<container-id> <image> <cidr> [name:addr...]")
+	}
+	containerID := args[0]
+	image := args[1]
+	cidrs := args[2]
+	extraHosts := args[3:]
+
+	container, err := inspectContainer(containerID)
+	if err != nil {
+		return err
+	}
+
+	hostsPath := container.HostsPath
+	fqdn := container.Config.Hostname + "." + container.Config.Domainname
+
+	var ips []*net.IPNet
+	for _, cidr := range strings.Fields(cidrs) {
+		_, ipnet, err := net.ParseCIDR(cidr)
+		if err != nil {
+			return err
+		}
+		ips = append(ips, ipnet)
+	}
+
+	docker := os.Getenv("DOCKER_HOST")
+	if docker == "" {
+		docker = "unix:///var/run/docker.sock"
+	}
+	p, err := proxy.StubProxy(proxy.Config{DockerHost: docker, Image: image})
+	if err != nil {
+		return err
+	}
+	return p.RewriteEtcHosts(hostsPath, fqdn, ips, extraHosts)
+}

prog/weaveutil/bridge.go

@@ -0,0 +1,57 @@
+package main
+
+import (
+	"fmt"
+	"strconv"
+
+	"github.com/weaveworks/weave/common"
+	weavenet "github.com/weaveworks/weave/net"
+)
+
+func detectBridgeType(args []string) error {
+	if len(args) != 2 {
+		cmdUsage("detect-bridge-type", "<weave-bridge-name> <datapath-name>")
+	}
+	bridgeType, err := weavenet.ExistingBridgeType(args[0], args[1])
+	if err != nil {
+		return err
+	} else if bridgeType == nil {
+		fmt.Println("none")
+	} else {
+		fmt.Println(bridgeType.String())
+	}
+	return nil
+}
+
+func createBridge(args []string) error {
+	if len(args) != 10 {
+		cmdUsage("create-bridge", "<docker-bridge-name> <weave-bridge-name> <datapath-name> <mtu> <port> <mac> <no-fastdp> <no-bridged-fastdp> <proc-path> <expect-npc>")
+	}
+
+	mtu, err := strconv.Atoi(args[3])
+	if err != nil && args[3] != "" {
+		return err
+	}
+	port, err := strconv.Atoi(args[4])
+	if err != nil {
+		return err
+	}
+	config := weavenet.BridgeConfig{
+		DockerBridgeName: args[0],
+		WeaveBridgeName:  args[1],
+		DatapathName:     args[2],
+		MTU:              mtu,
+		Port:             port,
+		Mac:              args[5],
+		NoFastdp:         args[6] != "",
+		NoBridgedFastdp:  args[7] != "",
+		NPC:              args[9] == "--expect-npc",
+	}
+	procPath := args[8]
+	bridgeType, err := weavenet.EnsureBridge(procPath, &config, common.Log)
+	if err != nil {
+		return err
+	}
+	fmt.Println(bridgeType.String())
+	return nil
+}

prog/weaveutil/container.go

@@ -0,0 +1,285 @@
+package main
+
+import (
+	"fmt"
+	"os"
+	"strings"
+
+	docker "github.com/fsouza/go-dockerclient"
+)
+
+func inspectContainer(containerNameOrID string) (*docker.Container, error) {
+	c, err := docker.NewVersionedClientFromEnv("1.18")
+	if err != nil {
+		return nil, fmt.Errorf("unable to connect to docker: %s", err)
+	}
+
+	container, err := c.InspectContainer(containerNameOrID)
+	if err != nil {
+		return nil, fmt.Errorf("unable to inspect container %s: %s", containerNameOrID, err)
+	}
+	return container, nil
+}
+
+func containerPid(containerID string) (int, error) {
+	container, err := inspectContainer(containerID)
+	if err != nil {
+		return 0, err
+
+	}
+
+	if container.State.Pid == 0 {
+		return 0, fmt.Errorf("container %s not running", containerID)
+	}
+	return container.State.Pid, nil
+}
+
+func containerID(args []string) error {
+	if len(args) < 1 {
+		cmdUsage("container-id", "<container-name-or-short-id>")
+	}
+
+	container, err := inspectContainer(args[0])
+	if err != nil {
+		return err
+	}
+
+	fmt.Print(container.ID)
+	return nil
+}
+
+func containerState(args []string) error {
+	if len(args) < 1 {
+		cmdUsage("container-state", "<container-id> [<image-prefix>]")
+	}
+
+	container, err := inspectContainer(args[0])
+	if err != nil {
+		return err
+	}
+
+	if len(args) == 1 {
+		fmt.Print(container.State.StateString())
+	} else {
+		imagePrefix := args[1]
+		if strings.HasPrefix(container.Config.Image, imagePrefix) {
+			fmt.Print(container.State.StateString())
+		} else {
+			if container.State.Running {
+				fmt.Print("running image mismatch: ", container.Config.Image)
+			} else {
+				fmt.Print("image mismatch: ", container.Config.Image)
+			}
+		}
+	}
+	return nil
+}
+
+func containerFQDN(args []string) error {
+	if len(args) < 1 {
+		cmdUsage("container-fqdn", "<container-id>")
+	}
+
+	container, err := inspectContainer(args[0])
+	if err != nil {
+		return err
+	}
+
+	fmt.Print(container.Config.Hostname, ".", container.Config.Domainname)
+	return nil
+}
+
+func runContainer(args []string) error {
+	env := []string{}
+	name := ""
+	net := ""
+	pid := ""
+	privileged := false
+	restart := docker.NeverRestart()
+	volumes := []string{}
+	volumesFrom := []string{}
+
+	done := false
+	for i := 0; i < len(args) && !done; {
+		switch args[i] {
+		case "-e", "--env":
+			if v := os.Getenv(args[i+1]); v != "" {
+				env = append(env, args[i+1]+"="+v)
+			} else {
+				env = append(env, args[i+1])
+			}
+			args = append(args[:i], args[i+2:]...)
+		case "--name":
+			name = args[i+1]
+			args = append(args[:i], args[i+2:]...)
+		case "--net":
+			net = args[i+1]
+			args = append(args[:i], args[i+2:]...)
+		case "--pid":
+			pid = args[i+1]
+			args = append(args[:i], args[i+2:]...)
+		case "--privileged":
+			privileged = true
+			args = append(args[:i], args[i+1:]...)
+		case "--restart":
+			restart = docker.RestartPolicy{Name: args[i+1]}
+			args = append(args[:i], args[i+2:]...)
+		case "-v", "--volume":
+			// Must dedup binds, otherwise, container fails creation
+			skip := false
+			for _, v := range volumes {
+				if v == args[i+1] {
+					skip = true
+				}
+			}
+			if !skip {
+				volumes = append(volumes, args[i+1])
+			}
+			args = append(args[:i], args[i+2:]...)
+		case "--volumes-from":
+			volumesFrom = append(volumesFrom, args[i+1])
+			args = append(args[:i], args[i+2:]...)
+		default:
+			done = true
+		}
+	}
+
+	if len(args) < 2 {
+		cmdUsage("run-container", `[options] <image> <cmd> [[<cmd-options>] <cmd-arg1> [<cmd-arg2> ...]]
+
+  -e, --env list                       Set environment variables
+      --name string                    Assign a name to the container
+      --net string                     Network Mode
+      --pid string                     PID namespace to use
+      --privileged                     Give extended privileges to this container
+      --restart string                 Restart policy to apply when a container exits (default "no")
+  -v, --volume list                    Bind mount a volume
+      --volumes-from list              Mount volumes from the specified container(s)
+`)
+	}
+
+	image := args[0]
+	cmds := args[1:]
+
+	c, err := docker.NewVersionedClientFromEnv("1.18")
+	if err != nil {
+		return fmt.Errorf("unable to connect to docker: %s", err)
+	}
+
+	config := docker.Config{Image: image, Env: env, Cmd: cmds}
+	hostConfig := docker.HostConfig{NetworkMode: net, PidMode: pid, Privileged: privileged, RestartPolicy: restart, Binds: volumes, VolumesFrom: volumesFrom}
+	container, err := c.CreateContainer(docker.CreateContainerOptions{Name: name, Config: &config, HostConfig: &hostConfig})
+	if err != nil {
+		return fmt.Errorf("unable to create container: %s", err)
+	}
+
+	err = c.StartContainer(container.ID, &hostConfig)
+	if err != nil {
+		return fmt.Errorf("unable to start container: %s", err)
+	}
+
+	fmt.Print(container.ID)
+	return nil
+}
+
+func listContainers(args []string) error {
+	if len(args) > 1 {
+		cmdUsage("list-containers", "[<label>]")
+	}
+
+	c, err := docker.NewVersionedClientFromEnv("1.18")
+	if err != nil {
+		return fmt.Errorf("unable to connect to docker: %s", err)
+	}
+
+	opts := docker.ListContainersOptions{All: true}
+	if len(args) == 1 {
+		label := args[0]
+		opts.Filters = map[string][]string{"label": {label}}
+	}
+
+	containers, err := c.ListContainers(opts)
+	if err != nil {
+		return fmt.Errorf("unable to list containers: %s", err)
+	}
+
+	for _, container := range containers {
+		fmt.Println(container.ID)
+	}
+	return nil
+}
+
+func stopContainer(args []string) error {
+	if len(args) < 1 {
+		cmdUsage("stop-container", "<container-id> [<container-id2> ...]")
+	}
+
+	c, err := docker.NewVersionedClientFromEnv("1.18")
+	if err != nil {
+		return fmt.Errorf("unable to connect to docker: %s", err)
+	}
+
+	for _, containerID := range args {
+		err = c.StopContainer(containerID, 10)
+		if err != nil {
+			return fmt.Errorf("unable to stop container %s: %s", containerID, err)
+		}
+	}
+	return nil
+}
+
+func killContainer(args []string) error {
+	if len(args) < 1 {
+		cmdUsage("kill-container", "<container-id> [<container-id2> ...]")
+	}
+
+	c, err := docker.NewVersionedClientFromEnv("1.18")
+	if err != nil {
+		return fmt.Errorf("unable to connect to docker: %s", err)
+	}
+
+	for _, containerID := range args {
+		err = c.KillContainer(docker.KillContainerOptions{ID: containerID, Signal: docker.SIGKILL})
+		if err != nil {
+			return fmt.Errorf("unable to stop container %s: %s", containerID, err)
+		}
+	}
+	return nil
+}
+
+func removeContainer(args []string) error {
+	if len(args) < 1 {
+		cmdUsage("remove-container", "[-f | --force]  [-v | --volumes] <container-id> [<container-id2> ...]")
+	}
+
+	force := false
+	volumes := false
+	for i := 0; i < len(args); {
+		switch args[i] {
+		case "--force":
+		case "-f":
+			force = true
+			args = append(args[:i], args[i+1:]...)
+		case "--volumes":
+		case "-v":
+			volumes = true
+			args = append(args[:i], args[i+1:]...)
+		default:
+			i++
+		}
+	}
+
+	c, err := docker.NewVersionedClientFromEnv("1.18")
+	if err != nil {
+		return fmt.Errorf("unable to connect to docker: %s", err)
+	}
+
+	for _, containerID := range args {
+		err = c.RemoveContainer(docker.RemoveContainerOptions{
+			ID: containerID, Force: force, RemoveVolumes: volumes})
+		if err != nil {
+			return fmt.Errorf("unable to stop container %s: %s", containerID, err)
+		}
+	}
+	return nil
+}

prog/weaveutil/db_flag.go

@@ -0,0 +1,54 @@
+// Get/set arbitrary flags in the Weave Net persistence DB
+package main
+
+import (
+	"fmt"
+
+	"github.com/weaveworks/weave/db"
+)
+
+const dbFlagPrefix = "flag:"
+
+func getDBFlag(args []string) error {
+	if len(args) != 2 {
+		cmdUsage("check-db-flag", "<db-prefix> <flag-name>")
+	}
+	dbPrefix := args[0]
+	flagName := args[1]
+
+	d, err := db.NewBoltDBReadOnly(dbPrefix)
+	if err != nil {
+		return err
+	}
+	defer d.Close()
+	var value string
+	nameFound, err := d.Load(dbFlagPrefix+flagName, &value)
+	if err != nil {
+		return err
+	}
+	if !nameFound {
+		return fmt.Errorf("")
+	}
+	fmt.Print(value)
+	return nil
+}
+
+func setDBFlag(args []string) error {
+	if len(args) != 3 {
+		cmdUsage("set-db-flag", "<db-prefix> <flag-name> <flag-value>")
+	}
+	dbPrefix := args[0]
+	flagName := args[1]
+	value := args[2]
+
+	d, err := db.NewBoltDB(dbPrefix)
+	if err != nil {
+		return err
+	}
+	defer d.Close()
+	err = d.Save(dbFlagPrefix+flagName, value)
+	if err != nil {
+		return err
+	}
+	return nil
+}

prog/weaveutil/expose.go

@@ -1,23 +0,0 @@
-package main
-
-import (
-	weavenet "github.com/weaveworks/weave/net"
-)
-
-func exposeNAT(args []string) error {
-	if len(args) < 1 {
-		cmdUsage("expose-nat", "<cidr>...")
-	}
-
-	cidrs, err := parseCIDRs(args)
-	if err != nil {
-		return err
-	}
-
-	for _, cidr := range cidrs {
-		if err := weavenet.ExposeNAT(*cidr); err != nil {
-			return err
-		}
-	}
-	return nil
-}

prog/weaveutil/main.go

@@ -14,29 +14,44 @@ var commands map[string]func([]string) error
 
 func init() {
 	commands = map[string]func([]string) error{
-		"help":                   help,
-		"netcheck":               netcheck,
-		"docker-tls-args":        dockerTLSArgs,
-		"create-datapath":        createDatapath,
-		"delete-datapath":        deleteDatapath,
-		"check-datapath":         checkDatapath,
-		"add-datapath-interface": addDatapathInterface,
-		"create-plugin-network":  createPluginNetwork,
-		"remove-plugin-network":  removePluginNetwork,
-		"container-addrs":        containerAddrs,
-		"process-addrs":          processAddrs,
-		"attach-container":       attach,
-		"detach-container":       detach,
-		"configure-arp":          configureARP,
-		"check-iface":            checkIface,
-		"del-iface":              delIface,
-		"setup-iface":            setupIface,
-		"list-netdevs":           listNetDevs,
-		"cni-net":                cniNet,
-		"cni-ipam":               cniIPAM,
-		"expose-nat":             exposeNAT,
-		"bridge-ip":              bridgeIP,
-		"unique-id":              uniqueID,
+		"help":                     help,
+		"netcheck":                 netcheck,
+		"docker-tls-args":          dockerTLSArgs,
+		"create-bridge":            createBridge,
+		"detect-bridge-type":       detectBridgeType,
+		"create-datapath":          createDatapath,
+		"delete-datapath":          deleteDatapath,
+		"check-datapath":           checkDatapath,
+		"add-datapath-interface":   addDatapathInterface,
+		"remove-plugin-network":    removePluginNetwork,
+		"container-addrs":          containerAddrs,
+		"process-addrs":            processAddrs,
+		"container-id":             containerID,
+		"container-state":          containerState,
+		"container-fqdn":           containerFQDN,
+		"run-container":            runContainer,
+		"list-containers":          listContainers,
+		"stop-container":           stopContainer,
+		"kill-container":           killContainer,
+		"remove-container":         removeContainer,
+		"create-volume-container":  createVolumeContainer,
+		"attach-container":         attach,
+		"detach-container":         detach,
+		"configure-arp":            configureARP,
+		"check-iface":              checkIface,
+		"del-iface":                delIface,
+		"setup-iface":              setupIface,
+		"setup-iface-addrs":        setupIfaceAddrs,
+		"list-netdevs":             listNetDevs,
+		"cni-net":                  cniNet,
+		"cni-ipam":                 cniIPAM,
+		"bridge-ip":                bridgeIP,
+		"unique-id":                uniqueID,
+		"swarm-manager-peers":      swarmManagerPeers,
+		"is-docker-plugin-enabled": isDockerPluginEnabled,
+		"rewrite-etc-hosts":        rewriteEtcHosts,
+		"get-db-flag":              getDBFlag,
+		"set-db-flag":              setDBFlag,
 	}
 }
 
@@ -65,11 +80,14 @@ func main() {
 	}
 	cmd, found := commands[os.Args[1]]
 	if !found {
+		fmt.Fprintf(os.Stderr, "%q cmd is not found\n", os.Args[1])
 		usage()
 		os.Exit(1)
 	}
 	if err := cmd(os.Args[2:]); err != nil {
-		fmt.Fprintln(os.Stderr, err)
+		if err.Error() != "" {
+			fmt.Fprintln(os.Stderr, err)
+		}
 		os.Exit(1)
 	}
 }

prog/weaveutil/netdev.go

@@ -6,7 +6,6 @@ import (
 	"strconv"
 	"strings"
 
-	"github.com/coreos/go-iptables/iptables"
 	"github.com/j-keck/arping"
 	"github.com/vishvananda/netlink"
 
@@ -48,33 +47,35 @@ func setupIface(args []string) error {
 	ifaceName := args[0]
 	newIfName := args[1]
 
-	ipt, err := iptables.New()
+	return weavenet.SetupIface(ifaceName, newIfName)
+}
+
+// setupIfaceAddrs sets up addresses on an interface. It expects to be called inside the container's netns.
+func setupIfaceAddrs(args []string) error {
+	if len(args) < 1 {
+		cmdUsage("setup-iface-addrs", "<iface-name> <with-multicast> <cidr>...")
+	}
+	link, err := netlink.LinkByName(args[0])
 	if err != nil {
 		return err
 	}
-
-	link, err := netlink.LinkByName(ifaceName)
+	withMulticastRoute, err := strconv.ParseBool(args[1])
 	if err != nil {
 		return err
 	}
-	if err := netlink.LinkSetName(link, newIfName); err != nil {
+	cidrs, err := parseCIDRs(args[2:])
+	if err != nil {
 		return err
 	}
-	if err := weavenet.ConfigureARPCache(newIfName); err != nil {
-		return err
-	}
-	if err := ipt.Append("filter", "INPUT", "-i", newIfName, "-d", "224.0.0.0/4", "-j", "DROP"); err != nil {
-		return err
-	}
-
-	return nil
+	return weavenet.SetupIfaceAddrs(link, withMulticastRoute, cidrs)
 }
 
 func configureARP(args []string) error {
-	if len(args) != 1 {
-		cmdUsage("configure-arp", "<iface-name-prefix>")
+	if len(args) != 2 {
+		cmdUsage("configure-arp", "<iface-name-prefix> <root-path>")
 	}
 	prefix := args[0]
+	rootPath := args[1]
 
 	links, err := netlink.LinkList()
 	if err != nil {
@@ -83,7 +84,7 @@ func configureARP(args []string) error {
 	for _, link := range links {
 		ifName := link.Attrs().Name
 		if strings.HasPrefix(ifName, prefix) {
-			weavenet.ConfigureARPCache(ifName)
+			weavenet.ConfigureARPCache(rootPath+"/proc", ifName)
 			if addrs, err := netlink.AddrList(link, netlink.FAMILY_V4); err == nil {
 				for _, addr := range addrs {
 					arping.GratuitousArpOverIfaceByName(addr.IPNet.IP, ifName)

prog/weaveutil/plugin_network.go

@@ -2,47 +2,14 @@
 package main
 
 import (
+	"context"
 	"fmt"
+	"strings"
 
+	"github.com/docker/docker/client"
 	docker "github.com/fsouza/go-dockerclient"
-
-	"github.com/weaveworks/weave/plugin/net"
 )
 
-func createPluginNetwork(args []string) error {
-	if len(args) != 3 {
-		cmdUsage("create-plugin-network", "<network-name> <driver-name> <default-subnet>")
-	}
-	networkName := args[0]
-	driverName := args[1]
-	subnet := args[2]
-	d, err := newDockerClient()
-	if err != nil {
-		return err
-	}
-	_, err = d.CreateNetwork(
-		docker.CreateNetworkOptions{
-			Name:           networkName,
-			CheckDuplicate: true,
-			Driver:         driverName,
-			IPAM: docker.IPAMOptions{
-				Driver: driverName,
-				Config: []docker.IPAMConfig{{Subnet: subnet}},
-			},
-			Options: map[string]interface{}{plugin.MulticastOption: "true"},
-		})
-	if err != docker.ErrNetworkAlreadyExists && err != nil {
-		// Despite appearances to the contrary, CreateNetwork does
-		// sometimes(always?) *not* return ErrNetworkAlreadyExists
-		// when the network already exists. Hence we need to check for
-		// this explicitly.
-		if _, err2 := d.NetworkInfo(networkName); err2 != nil {
-			return fmt.Errorf("unable to create network: %s", err)
-		}
-	}
-	return nil
-}
-
 func removePluginNetwork(args []string) error {
 	if len(args) != 1 {
 		cmdUsage("remove-plugin-network", "<network-name>")
@@ -71,6 +38,41 @@ while Weave is not running`, networkName, containers)
 	return nil
 }
 
+// Exits with 0 if the given plugin (v2) is enabled.
+//
+// Any failure due to missing plugin support is non-harmful as plugin (v2)
+// cannot be enabled when Docker does not support it.
+func isDockerPluginEnabled(args []string) error {
+	if len(args) != 1 {
+		cmdUsage("is-docker-plugin-enabled", "<plugin-name>")
+	}
+
+	pluginName := args[0]
+
+	// This is messed up: we are using docker/docker/client instead of
+	// fsouza/go-dockerclient because the latter does not support plugins.
+	c, err := client.NewEnvClient()
+	if err != nil {
+		return fmt.Errorf("unable to connect to docker: %s", err)
+	}
+
+	ctx := context.Background()
+
+	plugins, err := c.PluginList(ctx)
+	if err != nil {
+		return err
+	}
+
+	for _, p := range plugins {
+		if p.Enabled && strings.Contains(p.Name, pluginName) {
+			fmt.Println(p.Name)
+			return nil
+		}
+	}
+
+	return fmt.Errorf("plugin %q not found", pluginName)
+}
+
 func newDockerClient() (*docker.Client, error) {
 	// API 1.21 is the first version that supports docker network
 	// commands
@@ -82,5 +84,6 @@ func newDockerClient() (*docker.Client, error) {
 	if err != nil {
 		return nil, fmt.Errorf("unable to connect to docker: %s", err)
 	}
+
 	return c, nil
 }

prog/weaveutil/swarm.go

@@ -0,0 +1,55 @@
+package main
+
+import (
+	"fmt"
+	"strings"
+
+	"github.com/fsouza/go-dockerclient"
+	"github.com/pkg/errors"
+)
+
+const DOCKER_API_VERSION = "1.26"
+
+func swarmManagerPeers(args []string) error {
+	info, err := dockerInfo()
+	if err != nil {
+		return err
+	}
+
+	ips := make([]string, 0)
+
+	for _, managerNode := range info.Swarm.RemoteManagers {
+		ip, err := ipFromAddr(managerNode.Addr)
+		if err != nil {
+			return errors.Wrap(err, "ipFromAddr")
+		}
+		ips = append(ips, ip)
+	}
+
+	fmt.Println(strings.Join(ips, " "))
+
+	return nil
+}
+
+func dockerInfo() (*docker.DockerInfo, error) {
+	client, err := docker.NewVersionedClientFromEnv(DOCKER_API_VERSION)
+	if err != nil {
+		return nil, errors.Wrap(err, "docker.NewVersionedClientFromEnv")
+	}
+
+	info, err := client.Info()
+	if err != nil {
+		return nil, errors.Wrap(err, "docker.Info")
+	}
+
+	return info, nil
+}
+
+func ipFromAddr(addr string) (string, error) {
+	parts := strings.Split(addr, ":")
+	if len(parts) != 2 {
+		return "", fmt.Errorf("invalid address: %q", addr)
+	}
+
+	return parts[0], nil
+}

prog/weaveutil/volume.go

@@ -0,0 +1,42 @@
+package main
+
+import (
+	"fmt"
+
+	docker "github.com/fsouza/go-dockerclient"
+)
+
+func createVolumeContainer(args []string) error {
+	if len(args) < 3 {
+		cmdUsage("create-volume-container", "<name> <image-name-or-id> <label> [<bind-mount1> <bind-mount2> ...]")
+	}
+	containerName := args[0]
+	image := args[1]
+	label := args[2]
+	bindMounts := args[3:]
+
+	c, err := docker.NewVersionedClientFromEnv("1.18")
+	if err != nil {
+		return fmt.Errorf("unable to connect to docker: %s", err)
+	}
+
+	container, err := c.InspectContainer(containerName)
+	if err == nil {
+		fmt.Println("volume container already exists " + containerName + ":" + container.ID)
+		return nil
+	}
+
+	volumes := make(map[string]struct{})
+	for _, m := range bindMounts {
+		volumes[m] = struct{}{}
+	}
+	labels := map[string]string{label: ""}
+	config := docker.Config{Image: image, Volumes: volumes, Labels: labels, Entrypoint: []string{"data-only"}}
+	hostConfig := docker.HostConfig{}
+	_, err = c.CreateContainer(docker.CreateContainerOptions{Name: containerName, Config: &config, HostConfig: &hostConfig})
+	if err != nil {
+		return fmt.Errorf("unable to create container: %s", err)
+	}
+
+	return nil
+}

proxy/common.go

@@ -6,13 +6,12 @@ import (
 	"fmt"
 	"io/ioutil"
 	"net/http"
-	"os"
-	"os/exec"
 	"regexp"
 	"strings"
 
 	"github.com/fsouza/go-dockerclient"
 	"github.com/weaveworks/weave/common"
+	weavedocker "github.com/weaveworks/weave/common/docker"
 )
 
 var (
@@ -24,39 +23,6 @@ var (
 	Log = common.Log
 )
 
-func callWeave(args ...string) ([]byte, []byte, error) {
-	args = append([]string{"--local"}, args...)
-	cmd := exec.Command("./weave", args...)
-	cmd.Env = []string{
-		"PATH=/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
-	}
-
-	propagateEnv := func(key string) {
-		if val := os.Getenv(key); val != "" {
-			cmd.Env = append(cmd.Env, fmt.Sprintf("%s=%s", key, val))
-		}
-	}
-
-	propagateEnv("DOCKER_HOST")
-
-	// Propagate WEAVE_DEBUG, to make debugging easier.
-	propagateEnv("WEAVE_DEBUG")
-
-	// This prevents the code coverage contortions in our
-	// integration test suite breaking things.
-	propagateEnv("COVERAGE")
-
-	// In case the router control endpoint address is non-standard.
-	propagateEnv("WEAVE_HTTP_ADDR")
-
-	Log.Debug("Calling weave args: ", args, "env: ", cmd.Env)
-	var stdout, stderr bytes.Buffer
-	cmd.Stdout = &stdout
-	cmd.Stderr = &stderr
-	err := cmd.Run()
-	return stdout.Bytes(), stderr.Bytes(), err
-}
-
 func unmarshalRequestBody(r *http.Request, target interface{}) error {
 	body, err := ioutil.ReadAll(r.Body)
 	if err != nil {
@@ -113,7 +79,7 @@ func marshalResponseBody(r *http.Response, body interface{}) error {
 	return nil
 }
 
-func inspectContainerInPath(client *docker.Client, path string) (*docker.Container, error) {
+func inspectContainerInPath(client *weavedocker.Client, path string) (*docker.Container, error) {
 	subs := containerIDRegexp.FindStringSubmatch(path)
 	if subs == nil {
 		err := fmt.Errorf("No container id found in request with path %s", path)

proxy/hosts.go

@@ -0,0 +1,64 @@
+package proxy
+
+import (
+	"bytes"
+	"fmt"
+	"io"
+	"net"
+	"path"
+	"strings"
+)
+
+// rewrite /etc/hosts, unlinking the file (so Docker does not modify it again) but
+// leaving it with valid contents...
+func (proxy *Proxy) RewriteEtcHosts(hostsPath, fqdn string, ips []*net.IPNet, extraHosts []string) error {
+	hostsPathDir := path.Dir(hostsPath)
+	mnt := "/container"
+	mntHosts := path.Join(mnt, path.Base(hostsPath))
+	var buf bytes.Buffer
+	writeEtcHostsContents(&buf, fqdn, ips, extraHosts)
+	contents := buf.String()
+	cmdLine := fmt.Sprintf("echo '%s' > %s && rm -f %s && echo '%s' > %s", contents, mntHosts, mntHosts, contents, mntHosts)
+	mounts := []string{hostsPathDir + ":" + mnt}
+	return proxy.runTransientContainer([]string{"sh"}, []string{"-c", cmdLine}, mounts)
+}
+
+// we assume (for compatibility with the weave script) that fqdn has a dot
+// between hostname and domain. domain may be blank.
+func writeEtcHostsContents(w io.Writer, fqdn string, cidrs []*net.IPNet, extraHosts []string) {
+	index := strings.Index(fqdn, ".")
+	name := fqdn[:index]
+	var hostnames string
+	if index == len(fqdn)-1 { // dot comes at the end - no domain
+		hostnames = name
+	} else {
+		if fqdn[len(fqdn)-1] == '.' { // strip a final trailing dot
+			fqdn = fqdn[:len(fqdn)-1]
+		}
+		hostnames = fqdn + " " + name
+	}
+
+	fmt.Fprintln(w, "# created by Weave - BEGIN")
+	fmt.Fprintln(w, "# container hostname")
+	for _, cidr := range cidrs {
+		fmt.Fprintf(w, "%-15s %s\n", cidr.IP, hostnames)
+	}
+	fmt.Fprintln(w)
+	fmt.Fprintln(w, "# static names added with --add-host")
+	for _, eh := range extraHosts {
+		parts := strings.SplitN(eh, ":", 2)
+		if len(parts) != 2 {
+			continue
+		}
+		fmt.Fprintf(w, "%-15s %s\n", parts[1], parts[0])
+	}
+	fmt.Fprintln(w)
+	fmt.Fprintln(w, "# default localhost entries")
+	fmt.Fprintln(w, "127.0.0.1       localhost")
+	fmt.Fprintln(w, "::1             ip6-localhost ip6-loopback")
+	fmt.Fprintln(w, "fe00::0         ip6-localnet")
+	fmt.Fprintln(w, "ff00::0         ip6-mcastprefix")
+	fmt.Fprintln(w, "ff02::1         ip6-allnodes")
+	fmt.Fprintln(w, "ff02::2         ip6-allrouters")
+	fmt.Fprintln(w, "# created by Weave - END")
+}

proxy/proxy.go

@@ -3,7 +3,6 @@ package proxy
 import (
 	"bytes"
 	"crypto/tls"
-	"errors"
 	"fmt"
 	"net"
 	"net/http"
@@ -16,10 +15,11 @@ import (
 	"time"
 
 	docker "github.com/fsouza/go-dockerclient"
+	"github.com/pkg/errors"
+
 	weaveapi "github.com/weaveworks/weave/api"
 	weavedocker "github.com/weaveworks/weave/common/docker"
 	weavenet "github.com/weaveworks/weave/net"
-	"github.com/weaveworks/weave/net/address"
 )
 
 const (
@@ -50,6 +50,7 @@ func dockerAPIEndpoint(endpoint string) *regexp.Regexp {
 }
 
 type Config struct {
+	Enabled             bool
 	HostnameFromLabel   string
 	HostnameMatch       string
 	HostnameReplacement string
@@ -59,9 +60,9 @@ type Config struct {
 	NoDefaultIPAM       bool
 	NoRewriteHosts      bool
 	TLSConfig           TLSConfig
-	Version             string
 	WithoutDNS          bool
 	NoMulticastRoute    bool
+	KeepTXOn            bool
 	DockerBridge        string
 	DockerHost          string
 }
@@ -75,7 +76,8 @@ type wait struct {
 type Proxy struct {
 	sync.Mutex
 	Config
-	client                 *docker.Client
+	client                 *weavedocker.Client
+	weave                  *weaveapi.Client
 	dockerBridgeIP         string
 	hostnameMatchRegexp    *regexp.Regexp
 	weaveWaitVolume        string
@@ -122,16 +124,13 @@ func (j attachJob) Stop() {
 	j.timer.Stop()
 }
 
-func NewProxy(c Config) (*Proxy, error) {
+func StubProxy(c Config) (*Proxy, error) {
 	p := &Proxy{
 		Config:     c,
 		waiters:    make(map[*http.Request]*wait),
 		attachJobs: make(map[string]*attachJob),
 		quit:       make(chan struct{}),
-	}
-
-	if err := p.TLSConfig.LoadCerts(); err != nil {
-		Log.Fatalf("Could not configure tls for proxy: %s", err)
+		weave:      weaveapi.NewClient(os.Getenv("WEAVE_HTTP_ADDR"), Log),
 	}
 
 	// We pin the protocol version to 1.18 (which corresponds to
@@ -143,9 +142,21 @@ func NewProxy(c Config) (*Proxy, error) {
 	if err != nil {
 		return nil, err
 	}
-	Log.Info(client.Info())
+	p.client = client
+	return p, nil
+}
 
-	p.client = client.Client
+func NewProxy(c Config) (*Proxy, error) {
+	p, err := StubProxy(c)
+	if err != nil {
+		return nil, err
+	}
+
+	if err := p.TLSConfig.LoadCerts(); err != nil {
+		Log.Fatalf("Could not configure tls for proxy: %s", err)
+	}
+
+	Log.Info(p.client.Info())
 
 	if !p.WithoutDNS {
 		ip, err := weavenet.FindBridgeIP(c.DockerBridge, nil)
@@ -166,7 +177,7 @@ func NewProxy(c Config) (*Proxy, error) {
 		return nil, err
 	}
 
-	client.AddObserver(p)
+	p.client.AddObserver(p)
 
 	return p, nil
 }
@@ -204,7 +215,7 @@ func (proxy *Proxy) findWeaveWaitVolumes() error {
 }
 
 func (proxy *Proxy) findVolume(v string) (string, error) {
-	container, err := proxy.client.InspectContainer("weaveproxy")
+	container, err := proxy.client.InspectContainer("weave")
 	if err != nil {
 		return "", fmt.Errorf("Could not find the weavewait volume: %s", err)
 	}
@@ -279,13 +290,17 @@ func (proxy *Proxy) Listen() []net.Listener {
 	return listeners
 }
 
-func (proxy *Proxy) Serve(listeners []net.Listener) {
+func (proxy *Proxy) Serve(listeners []net.Listener, ready func()) {
 	errs := make(chan error)
 	for _, listener := range listeners {
 		go func(listener net.Listener) {
 			errs <- (&http.Server{Handler: proxy}).Serve(listener)
 		}(listener)
 	}
+	// It would be better if we could delay calling Done() until all
+	// the listeners are ready, but it doesn't seem to be possible to
+	// hook the right point in http.Server
+	ready()
 	for range listeners {
 		err := <-errs
 		if err != nil {
@@ -294,17 +309,6 @@ func (proxy *Proxy) Serve(listeners []net.Listener) {
 	}
 }
 
-func (proxy *Proxy) ListenAndServeStatus(socket string) {
-	listener, err := weavenet.ListenUnixSocket(socket)
-	if err != nil {
-		Log.Fatalf("ListenAndServeStatus failed: %s", err)
-	}
-	handler := http.HandlerFunc(proxy.StatusHTTP)
-	if err := (&http.Server{Handler: handler}).Serve(listener); err != nil {
-		Log.Fatalf("ListenAndServeStatus failed: %s", err)
-	}
-}
-
 func (proxy *Proxy) StatusHTTP(w http.ResponseWriter, r *http.Request) {
 	for _, addr := range proxy.normalisedAddrs {
 		fmt.Fprintln(w, addr)
@@ -325,11 +329,7 @@ func copyOwnerAndPermissions(from, to string) error {
 		return nil
 	}
 
-	if err = os.Chown(to, int(moreStat.Uid), int(moreStat.Gid)); err != nil {
-		return err
-	}
-
-	return nil
+	return os.Chown(to, int(moreStat.Uid), int(moreStat.Gid))
 }
 
 func (proxy *Proxy) listen(protoAndAddr string) (net.Listener, string, error) {
@@ -408,11 +408,6 @@ func containerShouldAttach(container *docker.Container) bool {
 	return len(container.Config.Entrypoint) > 0 && container.Config.Entrypoint[0] == weaveWaitEntrypoint[0]
 }
 
-func containerIsWeaveRouter(container *docker.Container) bool {
-	return container.Name == weaveContainerName &&
-		len(container.Config.Entrypoint) > 0 && container.Config.Entrypoint[0] == weaveEntrypoint
-}
-
 func (proxy *Proxy) createWait(r *http.Request, ident string) {
 	proxy.Lock()
 	proxy.waiters[r] = &wait{ident: ident, ch: make(chan error, 1)}
@@ -489,10 +484,6 @@ func (proxy *Proxy) attach(containerID string) error {
 		}
 		return nil
 	}
-	if containerIsWeaveRouter(container) {
-		Log.Infof("Attaching weave router container: %s", container.ID)
-		return callWeaveAttach(container, []string{"attach-router"})
-	}
 	if !containerShouldAttach(container) || !container.State.Running {
 		return nil
 	}
@@ -503,53 +494,80 @@ func (proxy *Proxy) attach(containerID string) error {
 		return nil
 	}
 	Log.Infof("Attaching container %s with WEAVE_CIDR \"%s\" to weave network", container.ID, strings.Join(cidrs, " "))
-	if err := validateCIDRs(cidrs); err != nil {
+	ips, err := proxy.allocateCIDRs(container.ID, cidrs)
+	if err != nil {
 		return err
 	}
 
-	args := []string{"attach"}
-	args = append(args, cidrs...)
+	fqdn := container.Config.Hostname + "." + container.Config.Domainname
 	if !proxy.NoRewriteHosts {
-		args = append(args, "--rewrite-hosts")
-
+		var extraHosts []string
 		if container.HostConfig != nil {
-			for _, eh := range container.HostConfig.ExtraHosts {
-				args = append(args, fmt.Sprintf("--add-host=%s", eh))
+			extraHosts = container.HostConfig.ExtraHosts
+		}
+		if err := proxy.RewriteEtcHosts(container.HostsPath, fqdn, ips, extraHosts); err != nil {
+			Log.Errorf("Error from rewriting container's /etc/hosts: %s", err)
+			// carry on since the original /etc/hosts is good enough in most cases
+		}
+	}
+
+	pid := container.State.Pid
+	// Passing 0 for mtu means it will be taken from the bridge
+	err = weavenet.AttachContainer(weavenet.NSPathByPid(pid), fmt.Sprint(pid), weavenet.VethName, weavenet.WeaveBridgeName, 0, !proxy.NoMulticastRoute, ips, proxy.KeepTXOn, true)
+	if err != nil {
+		return err
+	}
+
+	if !proxy.WithoutDNS {
+		for _, ip := range ips {
+			if err := proxy.weave.RegisterWithDNS(container.ID, fqdn, ip.IP.String()); err != nil {
+				return errors.Wrapf(err, "unable to register %s with weaveDNS: %s", container.ID, err)
 			}
 		}
 	}
-	if proxy.NoMulticastRoute {
-		args = append(args, "--no-multicast-route")
-	}
-	args = append(args, container.ID)
-	return callWeaveAttach(container, args)
+
+	return err
 }
 
-func callWeaveAttach(container *docker.Container, args []string) error {
-	if _, stderr, err := callWeave(args...); err != nil {
-		Log.Warningf("Attaching container %s to weave network failed: %s", container.ID, string(stderr))
-		return errors.New(string(stderr))
-	} else if len(stderr) > 0 {
-		Log.Warningf("Attaching container %s to weave network: %s", container.ID, string(stderr))
+func (proxy *Proxy) allocateCIDRs(containerID string, cidrs []string) ([]*net.IPNet, error) {
+	if len(cidrs) == 0 {
+		cidrs = []string{"net:default"}
 	}
-	return nil
-}
-
-func validateCIDRs(cidrs []string) error {
+	var ipnet *net.IPNet
+	var err error
+	var ipnets []*net.IPNet
 	for _, cidr := range cidrs {
-		if cidr == "net:default" {
-			continue
-		}
-		for _, prefix := range []string{"ip:", "net:", ""} {
-			if strings.HasPrefix(cidr, prefix) {
-				if _, err := address.ParseCIDR(strings.TrimPrefix(cidr, prefix)); err == nil {
-					break
-				}
-				return fmt.Errorf("invalid WEAVE_CIDR: %s", cidr)
+		switch {
+		case cidr == "net:default":
+			ipnet, err = proxy.weave.AllocateIP(containerID, true)
+		case strings.HasPrefix(cidr, "net:"):
+			var subnet *net.IPNet
+			_, subnet, err = net.ParseCIDR(strings.TrimPrefix(cidr, "net:"))
+			if err != nil {
+				break
 			}
+			ipnet, err = proxy.weave.AllocateIPInSubnet(containerID, subnet, true)
+		case strings.HasPrefix(cidr, "ip:"):
+			ipnet, err = proxy.claimCIDR(containerID, strings.TrimPrefix(cidr, "ip:"))
+		default:
+			ipnet, err = proxy.claimCIDR(containerID, cidr)
 		}
+		if err != nil {
+			return nil, errors.Wrapf(err, "for %q", cidr)
+		}
+		ipnets = append(ipnets, ipnet)
 	}
-	return nil
+	return ipnets, nil
+}
+
+func (proxy *Proxy) claimCIDR(containerID, cidr string) (*net.IPNet, error) {
+	ip, ipnet, err := net.ParseCIDR(cidr)
+	if err != nil {
+		return nil, err
+	}
+	ipnet.IP = ip // we want the specific IP plus the mask
+	err = proxy.weave.ClaimIP(containerID, ipnet, true)
+	return ipnet, err
 }
 
 func (proxy *Proxy) weaveCIDRs(networkMode string, env []string) ([]string, error) {
@@ -598,8 +616,7 @@ func (proxy *Proxy) getDNSDomain() string {
 	if proxy.WithoutDNS {
 		return ""
 	}
-	weave := weaveapi.NewClient(os.Getenv("WEAVE_HTTP_ADDR"), Log)
-	domain, _ := weave.DNSDomain()
+	domain, _ := proxy.weave.DNSDomain()
 	return domain
 }
 
@@ -638,7 +655,6 @@ func (proxy *Proxy) updateContainerNetworkSettings(container jsonObject) error {
 }
 
 func (proxy *Proxy) symlink(unixAddrs []string) (err error) {
-	var container *docker.Container
 	binds := []string{"/var/run/weave:/var/run/weave"}
 	froms := []string{}
 	for _, addr := range unixAddrs {
@@ -654,7 +670,10 @@ func (proxy *Proxy) symlink(unixAddrs []string) (err error) {
 	if len(froms) == 0 {
 		return
 	}
+	return proxy.runTransientContainer([]string{"/home/weave/symlink", weaveSock}, froms, binds)
+}
 
+func (proxy *Proxy) runTransientContainer(entrypoint, cmd, binds []string) (err error) {
 	env := []string{
 		"PATH=/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
 	}
@@ -663,14 +682,16 @@ func (proxy *Proxy) symlink(unixAddrs []string) (err error) {
 		env = append(env, fmt.Sprintf("%s=%s", "WEAVE_DEBUG", val))
 	}
 
+	Log.Debugf("Running image %q; entrypoint=%q; cmd=%q; binds=%q", proxy.Image, entrypoint, fmt.Sprintf("%.72s", cmd), binds)
+	var container *docker.Container
 	container, err = proxy.client.CreateContainer(docker.CreateContainerOptions{
 		Config: &docker.Config{
 			Image:      proxy.Image,
-			Entrypoint: []string{"/home/weave/symlink", weaveSock},
-			Cmd:        froms,
+			Entrypoint: entrypoint,
+			Cmd:        cmd,
 			Env:        env,
 		},
-		HostConfig: &docker.HostConfig{Binds: binds},
+		HostConfig: &docker.HostConfig{Binds: binds, NetworkMode: "none"},
 	})
 	if err != nil {
 		return

proxy/status.go

@@ -0,0 +1,15 @@
+package proxy
+
+type Status struct {
+	Addresses []string
+}
+
+func NewStatus(proxy *Proxy) *Status {
+	if proxy == nil {
+		return nil
+	}
+	status := &Status{
+		Addresses: proxy.normalisedAddrs,
+	}
+	return status
+}

router/fastdp.go

@@ -209,25 +209,25 @@ func (lock *fastDatapathLock) relock() {
 	}
 }
 
-// Bridge bits
+// InjectorConsumer bits
 
-type fastDatapathBridge struct {
+type fastDatapathInjectorConsumer struct {
 	*FastDatapath
 }
 
-func (fastdp *FastDatapath) Bridge() Bridge {
-	return fastDatapathBridge{fastdp}
+func (fastdp *FastDatapath) InjectorConsumer() InjectorConsumer {
+	return fastDatapathInjectorConsumer{fastdp}
 }
 
-func (fastdp fastDatapathBridge) Interface() *net.Interface {
+func (fastdp fastDatapathInjectorConsumer) Interface() *net.Interface {
 	return fastdp.iface
 }
 
-func (fastdp fastDatapathBridge) String() string {
+func (fastdp fastDatapathInjectorConsumer) String() string {
 	return fmt.Sprint(fastdp.iface.Name, " (via ODP)")
 }
 
-func (fastdp fastDatapathBridge) Stats() map[string]int {
+func (fastdp fastDatapathInjectorConsumer) Stats() map[string]int {
 	lock := fastdp.startLock()
 	defer lock.unlock()
 
@@ -238,12 +238,12 @@ func (fastdp fastDatapathBridge) Stats() map[string]int {
 
 var routerBridgePortID = bridgePortID{router: true}
 
-func (fastdp fastDatapathBridge) StartConsumingPackets(consumer BridgeConsumer) error {
+func (fastdp fastDatapathInjectorConsumer) StartConsumingPackets(consumer Consumer) error {
 	fastdp.lock.Lock()
 	defer fastdp.lock.Unlock()
 
 	if fastdp.sendToPort[routerBridgePortID] != nil {
-		return fmt.Errorf("FastDatapath already has a BridgeConsumer")
+		return fmt.Errorf("FastDatapath already has a Consumer")
 	}
 
 	// set up delivery to the weave router port on the bridge
@@ -257,7 +257,7 @@ func (fastdp fastDatapathBridge) StartConsumingPackets(consumer BridgeConsumer)
 	return nil
 }
 
-func (fastdp fastDatapathBridge) InjectPacket(key PacketKey) FlowOp {
+func (fastdp fastDatapathInjectorConsumer) InjectPacket(key PacketKey) FlowOp {
 	lock := fastdp.startLock()
 	defer lock.unlock()
 	return fastdp.bridge(routerBridgePortID, key, &lock)

router/http.go

@@ -5,7 +5,10 @@ import (
 	"net/http"
 
 	"github.com/gorilla/mux"
+
 	"github.com/weaveworks/weave/common"
+	"github.com/weaveworks/weave/net"
+	"github.com/weaveworks/weave/net/address"
 )
 
 func (router *NetworkRouter) HandleHTTP(muxRouter *mux.Router) {
@@ -26,4 +29,20 @@ func (router *NetworkRouter) HandleHTTP(muxRouter *mux.Router) {
 		router.ForgetConnections(r.Form["peer"])
 	})
 
+	muxRouter.Methods("POST").Path("/expose/{ip}/{prefixlen}").HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		vars := mux.Vars(r)
+		cidr, err := address.ParseCIDR(vars["ip"] + "/" + vars["prefixlen"])
+		if err != nil {
+			http.Error(w, fmt.Sprint("unable to parse ip addr: ", err.Error()), http.StatusBadRequest)
+			return
+		}
+
+		if err = net.Expose(router.BridgeConfig.WeaveBridgeName, cidr.IPNet(), router.BridgeConfig.AWSVPC, router.BridgeConfig.NPC); err != nil {
+			http.Error(w, fmt.Sprint("unable to expose: ", err.Error()), http.StatusInternalServerError)
+			return
+		}
+
+		w.WriteHeader(204)
+	})
+
 }

router/injector_consumer.go

@@ -5,13 +5,13 @@ import (
 )
 
 // Interface to packet handling on the local virtual bridge
-type Bridge interface {
+type InjectorConsumer interface {
 	// Inject a packet to be delivered locally
 	InjectPacket(PacketKey) FlowOp
 
 	// Start consuming packets from the bridge.  Injected packets
 	// should not be included.
-	StartConsumingPackets(BridgeConsumer) error
+	StartConsumingPackets(Consumer) error
 
 	Interface() *net.Interface
 	String() string
@@ -19,26 +19,26 @@ type Bridge interface {
 }
 
 // A function that determines how to handle locally captured packets.
-type BridgeConsumer func(PacketKey) FlowOp
+type Consumer func(PacketKey) FlowOp
 
-type NullBridge struct{}
+type NullInjectorConsumer struct{}
 
-func (NullBridge) InjectPacket(PacketKey) FlowOp {
+func (NullInjectorConsumer) InjectPacket(PacketKey) FlowOp {
 	return nil
 }
 
-func (NullBridge) StartConsumingPackets(BridgeConsumer) error {
+func (NullInjectorConsumer) StartConsumingPackets(Consumer) error {
 	return nil
 }
 
-func (NullBridge) Interface() *net.Interface {
+func (NullInjectorConsumer) Interface() *net.Interface {
 	return nil
 }
 
-func (NullBridge) String() string {
+func (NullInjectorConsumer) String() string {
 	return "no overlay bridge"
 }
 
-func (NullBridge) Stats() map[string]int {
+func (NullInjectorConsumer) Stats() map[string]int {
 	return nil
 }

router/network_router.go

@@ -11,6 +11,7 @@ import (
 
 	"github.com/weaveworks/weave/common"
 	"github.com/weaveworks/weave/db"
+	weavenet "github.com/weaveworks/weave/net"
 )
 
 const (
@@ -34,9 +35,9 @@ var (
 )
 
 type NetworkConfig struct {
-	BufSz         int
-	PacketLogging PacketLogging
-	Bridge        Bridge
+	BufSz            int
+	PacketLogging    PacketLogging
+	InjectorConsumer InjectorConsumer
 }
 
 type PacketLogging interface {
@@ -47,23 +48,24 @@ type PacketLogging interface {
 type NetworkRouter struct {
 	*mesh.Router
 	NetworkConfig
+	weavenet.BridgeConfig
 	Macs *MacCache
 	db   db.DB
 }
 
-func NewNetworkRouter(config mesh.Config, networkConfig NetworkConfig, name mesh.PeerName, nickName string, overlay NetworkOverlay, db db.DB) (*NetworkRouter, error) {
+func NewNetworkRouter(config mesh.Config, networkConfig NetworkConfig, bridgeConfig weavenet.BridgeConfig, name mesh.PeerName, nickName string, overlay NetworkOverlay, db db.DB) (*NetworkRouter, error) {
 	if overlay == nil {
 		overlay = NullNetworkOverlay{}
 	}
-	if networkConfig.Bridge == nil {
-		networkConfig.Bridge = NullBridge{}
+	if networkConfig.InjectorConsumer == nil {
+		networkConfig.InjectorConsumer = NullInjectorConsumer{}
 	}
 
 	meshRouter, err := mesh.NewRouter(config, name, nickName, overlay, common.LogLogger())
 	if err != nil {
 		return nil, err
 	}
-	router := &NetworkRouter{Router: meshRouter, NetworkConfig: networkConfig, db: db}
+	router := &NetworkRouter{Router: meshRouter, NetworkConfig: networkConfig, BridgeConfig: bridgeConfig, db: db}
 	router.Peers.OnInvalidateShortIDs(overlay.InvalidateShortIDs)
 	router.Routes.OnChange(overlay.InvalidateRoutes)
 	router.Macs = NewMacCache(macMaxAge,
@@ -77,8 +79,8 @@ func NewNetworkRouter(config mesh.Config, networkConfig NetworkConfig, name mesh
 // Start listening for TCP connections, locally captured packets, and
 // forwarded packets.
 func (router *NetworkRouter) Start() {
-	log.Println("Sniffing traffic on", router.Bridge)
-	checkFatal(router.Bridge.StartConsumingPackets(router.handleCapturedPacket))
+	log.Println("Sniffing traffic on", router.InjectorConsumer)
+	checkFatal(router.InjectorConsumer.StartConsumingPackets(router.handleCapturedPacket))
 	checkFatal(router.Overlay.(NetworkOverlay).StartConsumingPackets(router.Ourself.Peer, router.Peers, router.handleForwardedPacket))
 	router.Router.Start()
 }
@@ -156,7 +158,7 @@ func (router *NetworkRouter) handleForwardedPacket(key ForwardPacketKey) FlowOp
 	}
 
 	router.PacketLogging.LogForwardPacket("Injecting", key)
-	injectFop := router.Bridge.InjectPacket(key.PacketKey)
+	injectFop := router.InjectorConsumer.InjectPacket(key.PacketKey)
 	dstPeer := router.Macs.Lookup(dstMac)
 	if dstPeer == router.Ourself.Peer {
 		return injectFop