Commit Graph

26813 Commits

Author SHA1 Message Date
Russ Cox
e4b369e1d7 storage: clean up timer in cacheWatcher.add
In the e2e benchmarks, this timer is a significant source of garbage
and stale timers. Because the timer is not stopped after its use
in the select, it stays in the timer heap until it eventually fires
(5 seconds later). Under load, a lot of 5-second timers can pile up
before any start going away. The timer heap being large makes timer
operations take longer; the operations are O(log N) but N is still big.

The way to fix this in current versions of Go is to stop the underlying
timer explicitly, which this CL does for this one case.

There are many other places in the code that use the same idiom,
but those do not show up on profiles of the e2e server.
I am investigating changes for Go 1.7's runtime that would make
the old code behave like this new code transparently, so I don't
think it's worth updating any uses of the idiom that are not in
hot spots found with profiling.

Measuring 'LIST nodes' latency in milliseconds during e2e test
shows the benefit of this change.

Using Go 1.4.2:

BEFORE  p50: 148±7   p90: 328±19  p99: 513±29  n: 10
AFTER   p50: 151±8   p90: 339±19  p99: 479±20  n: 9

Using Go 1.6.0:

BEFORE  p50: 141±9   p90: 383±32  p99: 604±44  n: 11
AFTER   p50: 140±14  p90: 360±31  p99: 483±39  n: 10
2016-03-18 15:58:34 -04:00
Prashanth Balasubramanian
cee32c0f84 Feature ingress release 1.2 2016-03-18 12:55:47 -07:00
Jeff Grafton
70813d6bb5 Pin to godep v58 in Jenkins unit/integration 2016-03-18 12:45:12 -07:00
Daniel Wang
848c64a17b Correct a typo in env vars
I should have used `KUBE_GCE_MASTER_PROJECT` and `KUBE_GCE_MASTER_IMAGE`.
2016-03-18 11:51:43 -07:00
Joe Finney
4877c988c3 Merge pull request #23086 from wonderfly/ci_jobs_for_trusty
Add CI jobs to track the master and 1.2 branch with Trusty images
2016-03-18 10:31:27 -07:00
Mike Spreitzer
2712b77652 Change default k8s version to download to 1.2.0
Since it is now out!
2016-03-18 11:41:11 -04:00
k8s-merge-robot
ec5ad0f774 Merge pull request #23112 from wojtek-t/rename_raw_json_in_raw_extension
Auto commit by PR queue bot
2016-03-18 08:30:20 -07:00
k8s-merge-robot
d2fe575218 Merge pull request #23137 from sondabar/master
Auto commit by PR queue bot
2016-03-18 04:47:31 -07:00
Wojciech Tyczynski
89585237cd Rename RawJSON to Raw in runtime.RawExtension and add ContentType & ContentEncoding. 2016-03-18 12:35:27 +01:00
k8s-merge-robot
e353d7a75a Merge pull request #23183 from gmarek/master
Auto commit by PR queue bot
2016-03-18 04:31:59 -07:00
gmarek
529af0fb84 Change kubemark-5 to use a single 2-core node 2016-03-18 10:59:43 +01:00
Wojciech Tyczynski
ce9b2ab3e3 Remove old deep-copy generator. 2016-03-18 09:42:15 +01:00
k8s-merge-robot
6bd5ecc192 Merge pull request #23164 from david-mcmahon/update-licenses
Auto commit by PR queue bot
2016-03-18 00:26:47 -07:00
Cole Mickens
c184f7ed21 kube2sky readme: fix flags 2016-03-17 21:59:07 -07:00
AdoHe
6c84fe5a69 change throttling debug output log level 2016-03-18 00:19:53 -04:00
k8s-merge-robot
10204f8b31 Merge pull request #23167 from cjcullen/sshcheck
Auto commit by PR queue bot
2016-03-17 20:36:20 -07:00
k8s-merge-robot
d4912eafb5 Merge pull request #23157 from spxtr/fix-tags
Auto commit by PR queue bot
2016-03-17 19:59:36 -07:00
k8s-merge-robot
98bede7f2d Merge pull request #23110 from wojtek-t/migrate_to_new_deep_copy_generator
Auto commit by PR queue bot
2016-03-17 19:21:23 -07:00
CJ Cullen
5f3929f75d Add a SSHKey sync check to the master's healthz (when using SSHTunnels). 2016-03-17 19:14:25 -07:00
k8s-merge-robot
78adb885c4 Merge pull request #23143 from vishh/23113
Auto commit by PR queue bot
2016-03-17 17:59:25 -07:00
k8s-merge-robot
26eab4d69c Merge pull request #23094 from spxtr/more-jobs
Auto commit by PR queue bot
2016-03-17 17:23:41 -07:00
David McMahon
ede7bb6793 Add a --retry to curl to help avoid transient failures. 2016-03-17 17:16:40 -07:00
David McMahon
98de58dd28 Merge pull request #19424 from david-mcmahon/versioning
Add section on branched patch releases
2016-03-17 16:21:39 -07:00
k8s-merge-robot
35d7469815 Merge pull request #23129 from mikedanese/kube-service-affinity
Auto commit by PR queue bot
2016-03-17 16:11:53 -07:00
Daniel Wang
3a56b68cc5 Add CI jobs to track the master and 1.2 branch with Trusty images
Add two CI jobs that run e2e tests on GCE with Trusty images (on both the master
and the nodes) on the `master` and `release-1.2` branch. Both tests use the
latest Trusty build from its `HEAD`. We will use these jobs to guard
kubernetes-Trusty compatibility.

These jobs mimic the existing `kubernetes-e2e-gce` and `kubernetes-e2e-gce-1-2`,
which follow `kubernetes-build` and `kubernetes-build-1-2`, respectively.

Additionally, move all Trusty related jobs into one section in the config file.
2016-03-17 15:17:24 -07:00
k8s-merge-robot
ad7a0da5af Merge pull request #23089 from saad-ali/addZoneToErr
Auto commit by PR queue bot
2016-03-17 15:07:28 -07:00
Joe Finney
9ba006572a Always use long format in git tag. 2016-03-17 15:07:25 -07:00
nikhiljindal
9ab16395e9 running hack/build-ui.sh 2016-03-17 15:05:25 -07:00
nikhiljindal
09aff21db9 Fixing merge conflict in swagger-ui/index.html 2016-03-17 15:05:25 -07:00
Hongchao Deng
dd6487d0b7 use '/usr/bin/env bash' in update/verify-godep-licenses.sh 2016-03-17 14:55:21 -07:00
Isaac Hollander McCreery
929c1a2bb3 Merge pull request #23151 from spxtr/staging-job
Add parallel gke-staging job.
2016-03-17 14:43:17 -07:00
Joe Finney
eb273e5a4b Add parallel gke-staging job. 2016-03-17 14:39:31 -07:00
Yu-Ju Hong
deafa44d61 kubelet: send all recevied pods in one update
The kubelet sync loop relies on getting one update as the signal that the
specific source is ready. This change ensures that we don't send multiple
updates (ADD, UPDATE) for the first batch of pods. This is required to prevent
the cleanup routine from killing pods prematurely.
2016-03-17 14:24:35 -07:00
k8s-merge-robot
782ba437f1 Merge pull request #23003 from deads2k/no-proxy-cidr
Auto commit by PR queue bot
2016-03-17 14:16:11 -07:00
Vishnu kannan
516559022c 1. Make kubelet default to 10ms for CPU quota if limit < 10m for
backwards compat.
2. Update documentation to reflect minimum CPU limits.

Signed-off-by: Vishnu kannan <vishnuk@google.com>
2016-03-17 13:30:06 -07:00
Jordan Liggitt
ea8669661a Use versioned object when computing patch 2016-03-17 16:19:05 -04:00
k8s-merge-robot
a7f1466556 Merge pull request #19577 from caesarxuchao/expose-swagger-for-discovery-types
Auto commit by PR queue bot
2016-03-17 12:50:17 -07:00
k8s-merge-robot
cda4583984 Merge pull request #23035 from xinxiaogang/xnxin-master
Auto commit by PR queue bot
2016-03-17 12:05:56 -07:00
Joe Finney
f1c7bd0307 Merge pull request #23139 from spxtr/bump-1-2-gke
Bump timeouts for 1.2 gke jobs.
2016-03-17 11:30:14 -07:00
Joe Finney
5c3228320c Bump timeouts for 1.2 gke jobs. 2016-03-17 11:28:09 -07:00
k8s-merge-robot
1427a7cf81 Merge pull request #23135 from spxtr/fix-enormous
Auto commit by PR queue bot
2016-03-17 11:24:10 -07:00
sondabar
b1d1ae8c21 Avoid certificate invalid messages for vagrant with parallels provider 2016-03-17 18:51:30 +01:00
Joe Finney
3ecf76b897 Fix enormous-cluster config. 2016-03-17 10:22:28 -07:00
CJ Cullen
6cd8e5c0e6 Merge pull request #23133 from spxtr/fix-staging
gke-staging should point to staging.
2016-03-17 10:17:14 -07:00
Joe Finney
6bd52b2e7e gke-staging should point to staging. 2016-03-17 10:09:11 -07:00
Marcin Wielgus
873baadbe4 Merge pull request #23128 from mwielgus/inc-go-timeout
Increase kubernetes-test-go timeout to 50 min for 1.2 and master
2016-03-17 18:00:45 +01:00
Mike Danese
e21ebbcac4 give the kubernetes service client ip session affinity 2016-03-17 09:38:22 -07:00
Marcin Wielgus
f396ed51b3 Increase kubernetes-test-go timeout to 50 min for 1.2 and master 2016-03-17 17:16:29 +01:00
k8s-merge-robot
435bc35812 Merge pull request #23091 from nikhiljindal/namespaceSwaggerSpec
Auto commit by PR queue bot
2016-03-17 08:35:06 -07:00
k8s-merge-robot
d2ae98c0f0 Merge pull request #22972 from ramitsurana/patch-3
Auto commit by PR queue bot
2016-03-17 07:58:19 -07:00