Commit Graph

42458 Commits

Author SHA1 Message Date
Krzysztof Siedlecki
4bdf48eb2a adding support for kubemark --node-labels flag 2019-10-11 11:26:42 +02:00
Guoliang Wang
b9167334ee [migration phase 1] PodFitsHost as filter plugin 2019-10-11 17:05:04 +08:00
Kubernetes Prow Robot
c1a735c642 Merge pull request #83601 from ahg-g/ahg-migration-priority
Implemented taints and tolerations priority function as a Score plugin
2019-10-11 00:05:50 -07:00
Louise Daly
a353247d44 Fixed bug in TopologyManager with SingleNUMANode Policy
This patch fixes an issue where best-effort pods were not admitted
to the node if the single-numa-node policy was set.

This was because the Admit policy in single-numa-node policy does
not admit any pod where the hint is anything but single NUMA node. The 'best hint' in this case is {<set bits for num. Numa Nodes on machine>, true}
So on a machine with 2 NUMA nodes the best hint for a best-effort pod is {11,true} as best-effort pods have no Topology preferences.

The single-numa-node policy fails any pod with a not preferred hint OR a hint where > 1 bits are set, thus the above example resulting in termintaed pods with a Topology Affinity Error.

This is a short term fix for the single-numa-node policy, as there will be code refactoring for the 1.17 release.
2019-10-11 07:00:37 +01:00
Kubernetes Prow Robot
1f5d329b48 Merge pull request #81158 from zouyee/node-fix
fix unsafe JSON construction
2019-10-10 22:49:40 -07:00
Guoliang Wang
d3e8f17987 Add more tracing steps in generic_scheduler 2019-10-11 10:39:41 +08:00
draveness
ee4dec65b5 feat(scheduler): expose SharedInformerFactory to the framework handle 2019-10-11 10:23:21 +08:00
Kubernetes Prow Robot
4561b67971 Merge pull request #83697 from klueska/fix-single-numa-with-one-provider
Fixed bug in TopologyManager with SingleNUMANode Policy
2019-10-10 19:00:33 -07:00
Kubernetes Prow Robot
a5e6ac0a95 Merge pull request #83694 from ahg-g/ahg-adv-config
[migration phase 1] Make scheduler cache, volume binder and listers available when registering default plugins
2019-10-10 19:00:21 -07:00
Kubernetes Prow Robot
c41942b81d Merge pull request #83674 from liu-cong/pod-metrics
Adding per-pod metrics for scheduler.
2019-10-10 19:00:10 -07:00
RainbowMango
633bb52b49 Deal with auto-generated files.
Update bazel by hack/update-bazel.sh
2019-10-11 09:26:58 +08:00
RainbowMango
3b07393ea8 Refactor UT with testutil from k/k. 2019-10-11 09:25:17 +08:00
Abdullah Gharaibeh
7429b0fe86 Implemented taints and tolerations priority function as a Score plugin 2019-10-10 17:49:40 -04:00
Cong Liu
085852160a Add per-pod metrics for scheduler. 2019-10-10 16:58:21 -04:00
Kubernetes Prow Robot
3db6d3abcf Merge pull request #83551 from dims/move-external-facing-kubelet-apis-to-staging
Move external facing kubelet apis to staging
2019-10-10 13:41:36 -07:00
Abdullah Gharaibeh
456df97745 made scheduler cache and volume binder available when instantiating factories for default plugins 2019-10-10 16:21:29 -04:00
Connor Doyle
a598369e3c Gofmt. 2019-10-10 12:16:21 -07:00
Connor Doyle
a9203ebdcf Topology manager aligns pods of all QoS classes. 2019-10-10 12:16:21 -07:00
Kubernetes Prow Robot
71bebe4c13 Merge pull request #83704 from SataQiu/metrics-kubelet-20191010
Remove direct references to prometheus/testutil from kubelet/metrics
2019-10-10 12:07:34 -07:00
Kubernetes Prow Robot
4f1c5b8cac Merge pull request #81940 from carlory/fix-appserver
fix static check failures
2019-10-10 12:07:21 -07:00
John Strunk
55881f2fc4 Improve efficiency of csiMountMgr.GetAttributes
GetAttributes is called repeatedly while setting the fsGroup of a
volume. Previously, it recalculated whether SELinux was supported during
each call. This resulted in volume.SetVolumeOwnership taking a long
time, delaying pod startup for high file count volumes.

This change checks the SELinux status once, right after node publish,
allowing GetAttributes to simply build and return a struct.

Signed-off-by: John Strunk <jstrunk@redhat.com>
2019-10-10 13:51:50 -04:00
carlory
f6bb24129e fix static check failures 2019-10-10 22:59:09 +08:00
Kevin Taylor
cb8a7c1a4c Promote VolumeSubpathEnvExpansion feature gate to GA 2019-10-10 09:34:40 +01:00
Kubernetes Prow Robot
2d8b8703de Merge pull request #83686 from ahg-g/ahg-noop
Remove the no-op plugin
2019-10-10 01:12:53 -07:00
Zida
bc56cf45a5 Fix comment typos in prober 2019-10-10 14:53:08 +08:00
SataQiu
23a8be6e5f remove direct references to prometheus/testutil from kubelet/metrics 2019-10-10 12:56:28 +08:00
zouyee
a864fd2100 fix unsafe JSON construction
Signed-off-by: Zou Nengren <zouyee1989@gmail.com>
2019-10-10 09:44:54 +08:00
Kevin Klues
5501f542cd Fixed bug in TopologyManager with SingleNUMANode Policy
This patch fixes an issue in the TopologyManager that wouldn't allow
pods to be admitted if pods were launched with the SingleNUMANode policy
and any of the hint providers had no NUMA preferences.

This is due to 2 factors:

1) Any hint provider that passes back a `nil` as its hints, has its hint
automatically transformed into a single {11 true} hint before merging

2) We added a special casing for the SingleNumaNodePolicy() in the
TopologyManager that essentially turns these hints into a
{11 false} anytime a {11 true} is seen.

The current patch reworks this logic so the that TopologyManager can
tell the difference between a "don't care" hint and a true "{11 true}"
hint returned by the hint provider. Only true "{11 true}" hints will be
converted by the special casing for the SingleNumaNodePolicy(), while
"don't care" hints will not.

This is a short term fix for this issue until we do a larger refactoring
of this code for the 1.17 release.
2019-10-09 17:41:08 -07:00
Iman Tabrizian
451a535401 feat: feat: change the pod_preemption_victims to Histogram 2019-10-09 18:28:32 -04:00
Kubernetes Prow Robot
a5dc1fffb0 Merge pull request #83543 from yutedz/attach-resync-comment
Remove stale comment about resyncPeriod
2019-10-09 13:17:50 -07:00
Kubernetes Prow Robot
0ff761b14f Merge pull request #82927 from obitech/kube_proxy_strict_serializer
Enable strict serializer in kube-proxy
2019-10-09 11:05:50 -07:00
Abdullah Gharaibeh
36a0f8a03f remove the no-op plugin that we used as an example plugin in default registry 2019-10-09 12:30:08 -04:00
Kubernetes Prow Robot
b8b7c376ed Merge pull request #81263 from draveness/feature/update-scheduling-queue-with-options
feat: update scheduling queue with options
2019-10-09 08:59:51 -07:00
Guangming Wang
97e2576750 hostport return error when claim SCTP type port
revert last changes, just modify typos in unit test, and remove failed assert

sub count of sctp port when assert of len compare
2019-10-09 20:02:04 +08:00
draveness
9646afb1f5 feat: add pod initial/max backoff duration to config API 2019-10-09 19:48:28 +08:00
Kubernetes Prow Robot
4b002b3baa Merge pull request #82123 from xiaoanyunfei/cleanup/take-effect-stateofworld-hashmap
replace iteration with hashmap in *state_of_world
2019-10-09 02:17:50 -07:00
Kubernetes Prow Robot
5ce75a8734 Merge pull request #83625 from mrobson/log-cgroups
Add podCgroup to process kill events to allow for correlation
2019-10-09 00:56:01 -07:00
RainbowMango
6099d49046 Deal with auto-generated files.
- Update bazel by hack/update-bazel.sh
2019-10-09 15:12:21 +08:00
RainbowMango
debe2f7b43 Refactor TestRunningPodAndContainerCount with metrics testutil 2019-10-09 15:09:23 +08:00
Kubernetes Prow Robot
057609cf93 Merge pull request #83638 from wking/informatio-typo
pkg/apis/policy: Fix "informatio" -> "information" comment typo
2019-10-08 23:03:49 -07:00
Kubernetes Prow Robot
e62ed95ecd Merge pull request #83611 from RainbowMango/pr_refactor_scheduler_test_with_testutils
Refactor scheduler metric test with testutils
2019-10-08 20:00:25 -07:00
Kubernetes Prow Robot
ac9390627e Merge pull request #83536 from yutedz/del-volume-err
Log the error return from store.Delete
2019-10-08 19:59:50 -07:00
Kubernetes Prow Robot
0110db3a14 Merge pull request #83537 from ahg-g/ahg-metadata
Added PredicateMetadata and PriorityMetadata to framework's CycleState
2019-10-08 18:37:50 -07:00
Kubernetes Prow Robot
72d052a444 Merge pull request #81797 from yastij/move-metrics-util
move util/metrics to component-base
2019-10-08 17:08:05 -07:00
Kubernetes Prow Robot
d8a3fbd275 Merge pull request #83599 from robscott/iptables-probability
Making iptables probability more granular in kube-proxy.
2019-10-08 15:42:52 -07:00
Kubernetes Prow Robot
464952fcc8 Merge pull request #83286 from draveness/feature/refactor-predicate-metadata
feat(scheduler): refactor predicateMetadata into a collection of sub types
2019-10-08 15:42:42 -07:00
Kubernetes Prow Robot
b00f009316 Merge pull request #82996 from tnqn/endpointslice-deletion
Fix EndpointSliceController service deletion processing
2019-10-08 15:42:27 -07:00
Kubernetes Prow Robot
b4489d1709 Merge pull request #82865 from tnqn/endpointslice
Fix wrong comments and inaccurate logs in endpointslice_controller
2019-10-08 15:42:16 -07:00
Kubernetes Prow Robot
4cff1c3ea1 Merge pull request #81280 from yqwang-ms/yqwang/exitmsglost
Fix Container exit message lost due to FallbackToLogsOnError is not compatible with ContainerCannotRun
2019-10-08 15:41:51 -07:00
Dan Winship
ebe94f139a Don't log about deleting non-existent affinity entries in userspace OnEndpointsAdd 2019-10-08 18:17:41 -04:00