Bryan Boreham
15bdf6cd2d
Rework CNI test so it uses the scriptlet
2016-04-11 16:00:49 +01:00
Bryan Boreham
4bd6de5995
Extract out CNI setup so we can call it from a test
2016-04-11 15:28:11 +01:00
Bryan Boreham
737b0e0504
Two different configs; ensure weave-ipam driver works when named
2016-04-08 17:59:23 +01:00
Bryan Boreham
13f3ff15bd
Sanity-check ipam args to make diagnosis easier
2016-04-08 15:18:08 +01:00
Bryan Boreham
e88e67f02c
Bind-mount docker socket for CNI plugin scripts
2016-04-08 15:16:54 +01:00
Adam Harrison
9c7b27acf0
Merge pull request #2134 from weaveworks/issues/2063-observer-status
...
Better log and status messages for Paxos observer
2016-04-08 10:11:46 +01:00
Bryan Boreham
aad5b4dd7e
Document observer status
2016-04-07 13:16:15 +01:00
Bryan Boreham
c12a52d20e
Better log and status messages for Paxos observer
2016-04-07 12:50:52 +01:00
Bryan Boreham
0c59de40e7
Refactor Paxos status call to remove type assertion
2016-04-07 12:12:08 +01:00
Adam Harrison
a897a7711c
Merge pull request #2115 from weaveworks/issues/2114-docs-in-wordpress
...
Restructure `/site` to conform to Wordepress requirements
2016-04-07 11:22:40 +01:00
Adam Harrison
fb8167c6f1
Fix broken link
2016-04-07 11:04:35 +01:00
Adam Harrison
0482d0716f
Fix broken links
2016-04-07 10:56:56 +01:00
Adam Harrison
f183cb491f
Adjust section links
2016-04-07 10:51:47 +01:00
Adam Harrison
80770cc96e
Restructure docs
2016-04-07 10:37:57 +01:00
Adam Harrison
21d061fa44
Remove Jekyll/Travis config
2016-04-07 10:36:10 +01:00
Matthias Radestock
7d27df0907
Merge pull request #2133 from weaveworks/issues/2081-broadcast-dns-overrides
...
Immediately broadcast DNS records that "re-assert a peer's authority"
2016-04-06 14:51:22 +01:00
Bryan Boreham
768e2a31fa
Re-hide output coming from stop, since it's generally uninteresting
2016-04-06 13:54:12 +01:00
Bryan Boreham
487392b99c
Immediately broadcast DNS records that "re-assert a peer's authority"
...
and enable test to check.
2016-04-06 13:35:07 +01:00
Matthias Radestock
50af75972a
Merge pull request #2122 from weaveworks/issues/2081-gossip-to-restarted-peer
...
Update weaveworks/mesh to send gossip to restarted peer
Fixes #2085 .
2016-04-06 12:01:38 +01:00
Matthias Radestock
88518ba278
Merge pull request #2130 from weaveworks/issues/2108-reboot-docs
...
Update reboot docs
Fixes #2108 .
2016-04-06 10:30:38 +01:00
Bryan Boreham
a83359da7d
Update reboots section now we have --restart=always
2016-04-06 10:29:53 +01:00
Bryan Boreham
9c4b4ca35f
Remove recursive link
2016-04-06 10:02:57 +01:00
Matthias Radestock
2e31087818
dump state on paxos test failure
...
to help diagnosis of #2132 and similar problems
2016-04-06 08:55:49 +01:00
Matthias Radestock
11e520e212
Merge pull request #2129 from weaveworks/fix-cov-collection
...
Do not fail during teardown of test suites if $COVERAGE is not set
2016-04-05 22:17:40 +01:00
Matthias Radestock
3254870899
cosmetic: tweak proxy intro to make more sense
2016-04-05 21:38:10 +01:00
Matthias Radestock
51210045f7
=1 is better than =true
...
though they both do the same thing
2016-04-05 20:16:15 +01:00
Matthias Radestock
9e6ae4422e
tweak checkpoint description
2016-04-05 20:02:46 +01:00
Matthias Radestock
ce431dd73f
Merge pull request #2096 from weaveworks/issues/2052-http-errors
...
Report errors when weave script talks to weave component via http
Fixes #2052 .
2016-04-05 19:35:01 +01:00
Bryan Boreham
934ec2467d
Eliminate race from call_weave; call check_running only after curl fails
2016-04-05 17:54:18 +01:00
Bryan Boreham
ff4d0dae70
Keep going if an IPAM claim errors
2016-04-05 17:54:18 +01:00
Matthias Radestock
bac3d65ede
refactor: use a 'case' for greater clarity
2016-04-05 17:54:18 +01:00
Bryan Boreham
ecb180c0c5
Ignore 404 from IP address claim when IPAM is disabled
2016-04-05 17:54:18 +01:00
Bryan Boreham
f01d91459b
Make http_call send error messages to stderr and return error codes,
...
and fix up all uses of this function.
2016-04-05 17:54:18 +01:00
Matthias Radestock
3e07f03c87
Merge branch '1.4'
2016-04-05 17:26:01 +01:00
Matthias Radestock
28e0b14220
Merge pull request #2128 from weaveworks/issues/2109-skip-empty-json
...
Don't attempt to parse JSON payload if it is 0 bytes long.
Fixes #2109 .
2016-04-05 17:24:37 +01:00
Matthias Radestock
f82ef55c67
Merge branch '1.4'
2016-04-05 17:21:09 +01:00
Matthias Radestock
53ba2f0aff
rename test so it actually gets run
2016-04-05 17:07:07 +01:00
Matthias Radestock
45b36b378d
Merge branch '1.4'
2016-04-05 16:57:47 +01:00
Martynas Pumputis
41a5c6f523
Do not fail if $COVERAGE is not set
...
Also, stop weave in end_suite by using `weave stop` which does not
remove the weave container (#1939 ).
2016-04-05 16:53:01 +01:00
Matthias Radestock
4f1d87f893
Merge pull request #2126 from weaveworks/2090-delay-start
...
Delay notifying observers about 'start' events
Fixes #2090 . Partially addresses #1775 .
2016-04-05 16:48:20 +01:00
Bryan Boreham
66c92fe075
Don't attempt to parse JSON payload if it is 0 bytes long.
...
Also added a test to trigger this.
2016-04-05 16:47:11 +01:00
Bryan Boreham
ebc5dcb4bb
Delay notifying observers about 'start' events
...
...until the container has a pid, or we see the same container die.
This is to work around https://github.com/docker/docker/issues/18574
Fixes #2090 . Partially addresses #1775 .
2016-04-05 16:20:07 +01:00
Bryan Boreham
afff92c248
Merge pull request #2124 from weaveworks/idempotent-tombstone
...
make DNS tombstoning idempotent
2016-04-05 17:15:53 +02:00
Matthias Radestock
ead3518b4b
make DNS tombstoning idempotent
...
so we avoid the unnecessary gossip, and delayed convergence, that
otherwise ensues.
2016-04-04 20:34:01 +01:00
Bryan Boreham
c38dcb1d73
Adapt to logging changes in weaveworks/mesh
2016-04-04 16:01:58 +01:00
Bryan Boreham
00fe47ba62
Update weaveworks/mesh to send gossip to restarted peer
2016-04-04 15:29:36 +01:00
Bryan Boreham
b7d49cdeb7
Merge pull request #2117 from weaveworks/2113-wait-for-dp
...
wait for datapath device to be ready
2016-04-01 12:13:21 +02:00
Matthias Radestock
495a35a6a6
work around fd re-use bug in vishvananda/netlink
...
Closing a subscription can cause the associated fd to be re-used while
the subscription goroutine is also still trying to use it. Which can
have disastrous consequences.
So as a work around we don't close subscriptions. This means we leak
an fd, some buffer space, and goroutine. The latter will eventually
block in a channel send of a subscription event, since nothing is left
reading ont the other end. So at least it won't be consuming cycles.
2016-04-01 11:07:19 +01:00
Bryan Boreham
9f13d28508
Merge pull request #2116 from weaveworks/plugin-coverage
...
add coverage to plugin
2016-04-01 12:01:25 +02:00
Matthias Radestock
de4cc0ad54
eradicate last remaining wEaVeDnS misspellings
2016-03-31 06:39:41 +01:00
Bryan Boreham