Matthias Radestock
bfb050c9cc
Previously when a peer restarted, information about the new incarnation (i.e. with a different UID) was not accepted by other peers (and connections would be dropped) unless all knowledge of the previous incarnation had been purged. This could result in a lot of connection churn and hence connectivity disruption, and, in some pathological cases, very slow convergence and hence acceptance of the new incarnation into the network. We now no longer drop connections when encountering different incarnations of a peer. There are two situations when that can happen: 1) on connection establishment we simply proceed 2) on receipt of gossip to ensure convergence we a) treat the UID as an additional discriminator when deciding whether we should update our information about a peer with that which was gossiped. Specifically, we update the information we hold when a) the gossiped version is greater, or b) is the same and the UID is greater. b) include the UID in the information we update c) move our own version number beyond any we receive for ourselves, if the received UID differs from ours. With (a) we establishes a total order of peer information across several incarnations of the same peer. i.e. we consider information to be fresher if it has a higher version, or the same version and higher UID. This may seem somewhat counter intutive, since it will generally treat information about new incarnations as older than old incarnations, since incarnations always start life with version 1. But to do better we'd need to establish a total order of incarnations that matches their temporal occurrence. Which requires some sort of durable state. So instead we have (c). Through that we learn the highest version number of any old incarnation of ourselves that other peers still hold, and then make sure that our version is greater than that. Essentially we continue where the old incarnations left off. It's as if instead of restarting we had simply changed UIDs. And due to (a) and (b) the information about the new incarnation of ourselves, now with a higher version, will supersede that of the old incarnations. Fixes #1554.
Weave - weaving containers into applications
About Weaveworks
Weaveworks is the company that develops Weave - the most productive way for developers to connect, observe and control Docker containers. To learn about our products, including getting started tutorials, visit our website and documentation or continue to read about some of the more technical aspects of Weave in this readme.
Weave
Weave creates a virtual network that connects Docker containers deployed across multiple hosts and enables their automatic discovery.
Applications use the network just as if the containers were all plugged into the same network switch, with no need to configure port mappings, links, etc. Services provided by application containers on the weave network can be made accessible to the outside world, regardless of where those containers are running. Similarly, existing internal systems can be exposed to application containers irrespective of their location.
Weave can traverse firewalls and operate in partially connected networks. Traffic can be encrypted, allowing hosts to be connected across an untrusted network.
With weave you can easily construct applications consisting of multiple containers, running anywhere.
Weave works alongside Docker's existing (single host) networking capabilities, so these can continue to be used by containers.
Installation
Ensure you are running Linux (kernel 3.8 or later) and have Docker (version 1.3.1 or later) installed. Then install weave with
sudo curl -L git.io/weave -o /usr/local/bin/weave
sudo chmod a+x /usr/local/bin/weave
For usage on OSX (with Docker Machine) you first need to make sure that a VM is running and configured, as shown in the Docker Machine documentation. Then you can launch weave directly from the OSX host.
CoreOS users see here for an example of installing weave using cloud-config.
Amazon ECS users see here for the latest Weave AMIs and here to get started with Weave on ECS.
Weave respects the environment variable DOCKER_HOST, so you can run
it locally to control a weave network on a remote host.
Quick Start Screencast
Example
Say you have docker running on two hosts, accessible to each other as
$HOST1 and $HOST2, and want to deploy an application consisting of
two containers, one on each host.
On $HOST1 we run:
host1$ weave launch
host1$ eval $(weave env)
host1$ docker run --name a1 -ti ubuntu
NB: If the first command results in an error like
http:///var/run/docker.sock/v1.19/containers/create: dial unix /var/run/docker.sock: permission denied. Are you trying to connect to a TLS-enabled daemon without TLS?then you likely need to be 'root' in order to connect to the Docker daemon. If so, run the above and all subsequent commands in a single root shell (e.g. one created withsudo -s). Do not prefix individual commands withsudo, since some commands modify environment entries and hence they all need to be executed from the same shell.
The first line runs weave. The second line configures our environment so that containers launched via the docker command line are automatically attached to the weave network. Finally, we run our application container.
That's it! If our application consists of more than one container on
this host we simply launch them with docker run as appropriate.
Next we repeat similar steps on $HOST2...
host2$ weave launch $HOST1
host2$ eval $(weave env)
host2$ docker run --name a2 -ti ubuntu
The only difference, apart from the name of the application container,
is that we tell our weave that it should peer with the weave on
$HOST1 (specified as the IP address or hostname, and optional
:port, by which $HOST2 can reach it). NB: if there is a firewall
between $HOST1 and $HOST2, you must permit traffic to the weave
control port (TCP 6783) and data ports (UDP 6783/6784).
Note that we could instead have told the weave on $HOST1 to connect to
$HOST2, or told both about each other. Order does not matter here;
weave automatically (re)connects to peers when they become
available. Also, we can tell weave to connect to multiple peers by
supplying multiple addresses, separated by spaces. And we can
add peers dynamically.
Weave must be started once per host. The relevant container images are
pulled down on demand, but if you wish you can preload them by running
weave setup - this is particularly useful for automated deployments,
and ensures that there are no delays during later operations.
Now that we've got everything set up, let's see whether our containers can talk to each other...
In the container started on $HOST1...
root@a1:/# ping -c 1 -q a2
PING a2.weave.local (10.40.0.2) 56(84) bytes of data.
--- a2.weave.local ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.341/0.341/0.341/0.000 ms
Similarly, in the container started on $HOST2...
root@a2:/# ping -c 1 -q a1
PING a1.weave.local (10.32.0.2) 56(84) bytes of data.
--- a1.weave.local ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.366/0.366/0.366/0.000 ms
So there we have it, two containers on separate hosts happily talking to each other.
Find out more
Contact Us
Found a bug, want to suggest a feature, or have a question? You can
File an issue. And
there is a Weave Users Google
Group,
which you can post to by email at weave-users@weave.works. Or you can
contact Weaveworks directly at help@weave.works. When reporting a
bug, please include which version of weave you are running, as shown
by weave version.
Follow on Twitter: @weaveworks.
Read the Weave blog: Weaveblog.
IRC: #weavenetwork