reweave/weave
1
0 Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
350 Commits 7 Branches 85 Tags
4d5faafd0cae360cb1ad190389ccfebc9304450f
Commit Graph
350 Commits
This Branch
This Branch
All Branches
Author SHA1 Message Date
Matthias Radestock 4d5faafd0c refactor: order functions by dependency 2014-09-30 16:09:56 +01:00
Bryan Boreham a09a229b44 Improve robustness 2014-09-30 16:02:06 +01:00
Bryan Boreham c3fdee766f Add iptables rule to forward traffic across the bridge, to avoid ufw blocking it 2014-09-30 15:24:30 +01:00
Bryan Boreham a31d8c4d80 Refactor add_nat_rule into add_iptables_rule that is more general 2014-09-30 15:23:23 +01:00
Matthias Radestock 393b5cd459 automatically fetch dependencies 2014-09-30 13:13:41 +01:00
Bryan Boreham 4145528e24 Merge pull request #92 from dpw/87_multicast_route 
Route multicast packets across the weave network
 2014-09-30 12:50:47 +01:00
David Wragg 1366c01f6a Restore comment that got lost when rearranging code 2014-09-29 09:59:26 +01:00
David Wragg cb0541078a Route multicast packets across the weave network 
Add a route to send multicast packets to ethwe and so across the weave
network.

Without that, multicast packets get sent to the default route, i.e.
to docker's eth0 interface, rather than across the weave network.
So IP multicast doesn't work between containers on different hosts,
even though weaver supports it.
 2014-09-26 09:30:46 +01:00
David Wragg 28e74c166a Restructure attach and detach 
To make the control flow more obvious.
 2014-09-26 09:30:46 +01:00
Matthias Radestock 3a9b611400 perform a 'refresh' on cm.status 
so that we don't end up reporting targets under 'reconnects' that we
are already connected to.
 2014-09-26 06:44:03 +01:00
rade 6135a911c1 Merge pull request #89 from squaremo/proofread-readme 
Fixed various typos/cutpasta
 2014-09-25 18:02:28 +01:00
Michael Bridgen 84bfbcadb9 Fixed various typos/cutpasta 2014-09-25 17:51:37 +01:00
Bryan Boreham 707041c7d0 Merge branch 'master' of https://github.com/zettio/weave 2014-09-25 13:55:44 +01:00
Bryan Boreham e18950ccd0 Add note in the example about opening port 6783 2014-09-25 13:52:06 +01:00
rade ca2854e059 Merge pull request #83 from binocarlos/master 
correct weave usage output: change -passwd to -password
 2014-09-25 07:31:18 +01:00
Kai Davenport f6b8e9a675 change -passwd to -password in usage function 2014-09-24 14:19:40 -07:00
Bryan Boreham 573ec22087 Fixed typo 2014-09-24 12:43:56 +01:00
Matthias Radestock ada941cfcb Merge branch 'bboreham-54_reboot' 
Closes #54
 2014-09-24 12:24:31 +01:00
Matthias Radestock ebe8207f8a typo 2014-09-24 12:23:51 +01:00
Matthias Radestock 976c62a104 Merge branch 'bboreham-67_attach' 
Closes #67
 2014-09-24 12:20:47 +01:00
Matthias Radestock e64551c608 Merge branch '67_attach' of https://github.com/bboreham/weave into bboreham-67_attach 2014-09-24 11:41:28 +01:00
Bryan Boreham 704c3dad67 Improved comment 2014-09-24 10:17:45 +01:00
Bryan Boreham ad3a36686f Cosmetic 2014-09-24 10:11:36 +01:00
Bryan Boreham 57ad2cbeaf Merge remote-tracking branch 'upstream/master' into 67_attach 
Conflicts:
	pcap.go
 2014-09-24 10:09:25 +01:00
Bryan Boreham a4ffb4e45a Revert "don't set pcap timeout, to prevent cpu churn" 
This reverts commit b5df59660f.
 2014-09-24 10:07:49 +01:00
Bryan Boreham d5acb312b4 Rename 'configure_container_networking' to 'attach' 2014-09-24 10:01:52 +01:00
Bryan Boreham dc611fa0af Refactor common code in attach/detach 2014-09-24 09:39:44 +01:00
Matthias Radestock 53d4d469bb Merge branch 'master' into bboreham-67_attach 2014-09-23 18:09:46 +01:00
Matthias Radestock 025315363d prevent CPU churn by setting long pcap timeout 
pcap timeouts are a mystery. Setting the timeout to -1 used to reduce
latency, but that was before we started setting immediate mode. And it
turns out that with that timeout, pcap chews through a lot of CPU
polling its file descriptor, checking timers, etc. However, the
behaviour of setting *no* timeout is defined as, err, undefined. So
instead we set a very long timeout indeed. And loop if it expires.

Fixes #80
 2014-09-23 17:54:40 +01:00
Matthias Radestock d10ce1e801 reword 2014-09-23 17:24:52 +01:00
Bryan Boreham d80a5c8dff Revert "don't set pcap timeout, to prevent cpu churn" 
This reverts commit b5df59660f.
See issue #81
 2014-09-23 16:56:08 +01:00
Matthias Radestock 02efccc36a cosmetic 2014-09-23 16:18:00 +01:00
Matthias Radestock 5477f671db Merge branch '54_reboot' of https://github.com/bboreham/weave into bboreham-54_reboot 2014-09-23 16:13:45 +01:00
Bryan Boreham a21860689a Merge remote-tracking branch 'upstream/bboreham-67_attach' into 67_attach 2014-09-23 16:05:45 +01:00
Bryan Boreham 9e3cbba275 Capture error output from ip command and check if error was caused by container dying 2014-09-23 16:00:03 +01:00
Matthias Radestock ff3fbaede9 tweak & move docs for 'attach' and 'detach' 2014-09-23 15:32:53 +01:00
Matthias Radestock 8fa24fd6d4 Merge branch '67_attach' of https://github.com/bboreham/weave into bboreham-67_attach 2014-09-23 14:51:11 +01:00
Matthias Radestock f59b9577c8 Merge branch '67_attach' of https://github.com/bboreham/weave into bboreham-67_attach 2014-09-23 14:49:02 +01:00
Bryan Boreham 8e50e0759d Make attach and detach idempotent 2014-09-23 14:49:00 +01:00
Bryan Boreham 2d8318b158 Improved test for blank string 2014-09-23 13:53:20 +01:00
Bryan Boreham a6b83a1234 Updated instructions for attach and detach 2014-09-23 13:52:41 +01:00
Bryan Boreham d1896e86d0 Redesigned attach/detach 2014-09-23 13:01:21 +01:00
Matthias Radestock b5df59660f don't set pcap timeout, to prevent cpu churn 
pcap timeouts are a mystery. Setting the timeout to -1 used to reduce
latency, but that was before we started setting immediate mode. And it
turns out that with that timeout, pcap chews through a lot of CPU
polling its file descriptor, checking timers, etc.

With no timeout set an idle weave is completely idle - no system calls
at all.

Fixes #80
 2014-09-23 11:06:41 +01:00
Bryan Boreham 7815446032 Merge pull request #78 from zettio/77_eliminate_connection_maker_tick 
eliminate periodic 'tick' in connection_maker
 2014-09-23 10:58:25 +01:00
Matthias Radestock eecaa567fc eliminate periodic 'tick' in connection_maker 
checkStateAndAttemptConnections already performs a full scan of all
potential targets, from which we can easily determine the nearest
future time (or, rather, its duration from now) on which an attempt is
due.

We also need to kick the ConnectionMaker into action when we receive a
topology update containing new connections.
 2014-09-22 23:15:54 +01:00
Matthias Radestock d5e797f695 refactor: use better 'switch' 2014-09-22 22:34:46 +01:00
Matthias Radestock f11f84fe17 refactor: we don't care about the tick time 2014-09-22 19:38:21 +01:00
Matthias Radestock aa1e1b1d6c cosmetic - remove comment for which we have an issue filed 2014-09-22 18:34:50 +01:00
Matthias Radestock 9e8260dc63 cosmetic 2014-09-22 18:14:37 +01:00
rade e6041e082f Merge pull request #76 from bboreham/tag_fix 
Only push latest and git commit tag to DockerHub
 2014-09-22 18:04:40 +01:00