Commit Graph

23581 Commits

Author SHA1 Message Date
Kubernetes Prow Robot
2ee01fb673
Merge pull request #118945 from andrewsykim/update-request-filter-duration-buckets
Introduce larger buckets for request_filter_duration_seconds and request_wait_duration_seconds
2023-06-29 03:51:36 -07:00
Davanum Srinivas
931456a142
Simplify the node name for metrics - just use localhost
Signed-off-by: Davanum Srinivas <davanum@gmail.com>
2023-06-28 18:21:12 -04:00
Kubernetes Prow Robot
2190775b69
Merge pull request #118280 from stlaz/e2e_psa_labels
Set all PSa labels in tests
2023-06-28 11:14:43 -07:00
Andrew Sy Kim
16fdd3e5ee apiserver: introduce larger buckets for flowcontrol request_wait_duration_seconds bucket
Signed-off-by: Andrew Sy Kim <andrewsy@google.com>
2023-06-28 15:48:39 +00:00
Andrew Sy Kim
c607984379 apiserver: introduce larger buckets for request_filter_duration_seconds metric
Signed-off-by: Andrew Sy Kim <andrewsy@google.com>
2023-06-28 14:19:51 +00:00
Kubernetes Prow Robot
c78204dc06
Merge pull request #118202 from pohly/scheduler-perf-unit-test
scheduler-perf: run as integration tests
2023-06-28 06:24:31 -07:00
Kubernetes Prow Robot
ddbf3575a7
Merge pull request #116729 from AxeZhan/handlers_sync
[Scheduler] Make sure handlers have synced before scheduling
2023-06-28 01:26:31 -07:00
Patrick Ohly
0d41d509d2 scheduler_perf: replace gomega.Eventually with wait.PollUntilContextTimeout
This is done for the sake of consistency. The failure message becomes less
useful.
2023-06-28 09:22:26 +02:00
Patrick Ohly
cecebe8ea2 scheduler_perf: add TestScheduling integration test
This runs workloads that are labeled as "integration-test". The apiserver and
scheduler are only started once per unique configuration, followed by each
workload using that configuration. This makes execution faster. In contrast to
benchmarking, we care less about starting with a clean slate for each test.
2023-06-28 09:22:25 +02:00
Patrick Ohly
dfd646e0a8 scheduler_perf: fix namespace deletion
Merely deleting the namespace is not enough:
- Workloads might rely on the garbage collector to get rid of obsolete objects,
  so we should run it to be on the safe side.
- Pods must be force-deleted because kubelet is not running.
- Finally, the namespace controller is needed to get rid of
  deleted namespaces.
2023-06-28 09:22:25 +02:00
Patrick Ohly
d9c16a1ced scheduler_perf: fix goroutine leak in runWorkload
This becomes relevant when doing more fine-grained leak checking.
2023-06-28 08:14:34 +02:00
Patrick Ohly
2e7f37353c test/integration: avoid errors in fake PC controller during shutdown
Once the context is canceled, the controller can stop processing
events. Without this change it prints errors when the apiserver is already
down.
2023-06-28 08:14:34 +02:00
Kubernetes Prow Robot
960830bc66
Merge pull request #118102 from RomanBednar/retro-sc-assignment-ga
graduate RetroactiveDefaultStorageClass feature to GA in 1.28
2023-06-27 20:46:32 -07:00
Kubernetes Prow Robot
470889278f
Merge pull request #118910 from dims/better-url-for-scraping-metrics-from-kubelet
Better URL for scraping metrics from kubelet in node e2e tests
2023-06-27 16:00:41 -07:00
Kubernetes Prow Robot
48a6fb0c42
Merge pull request #118909 from dims/bump-to-latest-node-problem-detector-version-with-arm64
Bump to latest node-problem-detector version with arm64
2023-06-27 14:58:42 -07:00
Kubernetes Prow Robot
e4b3b4f20c
Merge pull request #118904 from dims/cleanup-pods-at-the-end-in-pod-conditions-e2e-node-test
Cleanup pods at the end in Pod conditions e2e node test
2023-06-27 14:58:31 -07:00
Davanum Srinivas
a75b00ea39
Better URL for scraping metrics from kubelet
Signed-off-by: Davanum Srinivas <davanum@gmail.com>
2023-06-27 16:14:59 -04:00
Davanum Srinivas
685b0c5efa
Bump to latest node-problem-detector version with arm64
Signed-off-by: Davanum Srinivas <davanum@gmail.com>
2023-06-27 16:04:12 -04:00
Kubernetes Prow Robot
b3d94ae74f
Merge pull request #118786 from pohly/dra-test-skip-prepare
dra: kubelet must skip NodePrepareResource if not used by any container
2023-06-27 09:58:32 -07:00
Kubernetes Prow Robot
76b2198da1
Merge pull request #118901 from dims/set-aws-specific-credential-provider-when-running-there
Set AWS specific credential provider when running there
2023-06-27 08:56:51 -07:00
Aldo Culquicondor
a4519665fe
Skip terminal Pods with a deletion timestamp from the Daemonset sync (#118716)
* Skip terminal Pods with a deletion timestamp from the Daemonset sync

Change-Id: I64a347a87c02ee2bd48be10e6fff380c8c81f742

* Review comments and fix integration test

Change-Id: I3eb5ec62bce8b4b150726a1e9b2b517c4e993713

* Include deleted terminal pods in history

Change-Id: I8b921157e6be1c809dd59f8035ec259ea4d96301
2023-06-27 08:56:33 -07:00
kidddddddddddddddddddddd
9c7166ff63 wait for eventhandlers to sync before run scheduler 2023-06-27 23:19:34 +08:00
Kubernetes Prow Robot
6dbb1c6cf0
Merge pull request #118902 from pohly/staticcheck-cleanup
Cleanup staticcheck workarounds, improve gomega calls, update to golangci-lint 1.53.3
2023-06-27 07:54:32 -07:00
Davanum Srinivas
6c587b43e9
Cleanup pods at the end in Pod conditions e2e node test
Signed-off-by: Davanum Srinivas <davanum@gmail.com>
2023-06-27 09:36:00 -04:00
Kevin Hannon
ecd727e4c7
Fix PodGC test when PodDisruptionConditions disabled (#118805)
* test comment should match the code in podgc

* Update test/integration/podgc/podgc_test.go

Co-authored-by: Michał Woźniak <mimowo@users.noreply.github.com>

* test comment should match the code in podgc

---------

Co-authored-by: Michał Woźniak <mimowo@users.noreply.github.com>
2023-06-27 05:50:29 -07:00
Madhav Jivrajani
bdbf07525f test: remove exception comments in discovery tests
The exception comments were added due to a false positive in
staticcheck. This has since been rectified.

Signed-off-by: Madhav Jivrajani <madhav.jiv@gmail.com>
2023-06-27 14:20:41 +02:00
Patrick Ohly
f618e61332 e2e: replace gomega.HaveLen(0) with gomega.BeEmpty()
The failure message becomes nicer. Found with the new ginkgolinter, for
example:

    test/e2e/apps/cronjob.go:113:3: ginkgo-linter: wrong length assertion; consider using `gomega.Expect(jobs.Items).To(gomega.BeEmpty())` instead (ginkgolinter)
     		gomega.Expect(jobs.Items).To(gomega.HaveLen(0))
     		^
2023-06-27 14:20:41 +02:00
Patrick Ohly
e7df337eba e2e: replace gomega.Equal(true/false) with gomega.BeTrue/BeFalse()
The failure message becomes a bit nicer. Found by the new ginkgolinter, for
example:

    test/e2e/windows/memory_limits.go:160:2: ginkgo-linter: wrong boolean assertion; consider using `gomega.Eventually(ctx, func() bool {
     eventList, err := f.ClientSet.CoreV1().Events(f.Namespace.Name).List(ctx, metav1.ListOptions{})
     ...
    }, 3*time.Minute, 10*time.Second).Should(gomega.BeTrue())` instead (ginkgolinter)
2023-06-27 14:20:20 +02:00
Patrick Ohly
8b33e8bdd1 e2e: fix gomega.Expect calls without assertions
"gomega.Expect" is not the same as "assert" in C: it always has to be combined
with a statement of what is expected.

Found with the new ginkgolinter, for example:

    test/e2e/node/pod_resize.go:242:3: ginkgo-linter: "Expect": missing assertion method. Expected "Should()", "To()", "ShouldNot()", "ToNot()" or "NotTo()" (ginkgolinter)
              gomega.Expect(found == true)
2023-06-27 14:20:20 +02:00
Davanum Srinivas
0ef1f2f2d8
Set AWS specific credential provider when running there
NOTE: we are not installing the ecr-credential-provider binary
itself here we are, we need to do it out-of-band from the test
suite itself before it runs.

Signed-off-by: Davanum Srinivas <davanum@gmail.com>
2023-06-27 07:45:59 -04:00
Kubernetes Prow Robot
9d50c0a025
Merge pull request #118889 from swetharepakula/extend-timeout
Extend Large Create Timeout to 60 min
2023-06-26 20:04:42 -07:00
Swetha Repakula
08198fe22b Extend Large Create Timeout to 60 min
* Load balancers are taking greater than 45 to provision
 * https://github.com/kubernetes/kubernetes/issues/118295
2023-06-26 15:32:34 -07:00
Dr. Stefan Schimanski
764da8a01d
FIXUP: cmd/kube-apiserver/app/options: split apart controlplane part 2023-06-26 21:50:38 +02:00
Kubernetes Prow Robot
9f5a3f5e90
Merge pull request #118868 from sttts/sttts-kube-apiserver-options-completion-move
cmd/kube-apiserver: move options completion into options package
2023-06-26 08:14:29 -07:00
Kubernetes Prow Robot
d9714078f8
Merge pull request #118551 from sanposhiho/event-to-register
feature(scheduler): implement ClusterEventWithHint to filter out useless events
2023-06-26 06:41:45 -07:00
Dr. Stefan Schimanski
6e079545c4
cmd/kube-apiserver: move options completion into options package 2023-06-26 15:20:40 +02:00
Kubernetes Prow Robot
a374d893f0
Merge pull request #118840 from AhmedGrati/fix-deployment-patch-e2e-test
fix: update deployment e2e tests
2023-06-26 04:49:44 -07:00
AhmedGrati
89a9e480f9 fix: update deployment e2e tests
Previously, we were trying to patch the deployment's ready replicas by
changing it to 0, at the same time we have 2 available replicas.
Therefore, this test fails since the ready replicas is less than the
available replicas. As a fix, we make the number of ready replicas equals to the number
of available ones.

Signed-off-by: AhmedGrati <ahmedgrati1999@gmail.com>
2023-06-24 20:48:01 +01:00
Kubernetes Prow Robot
e81c24dfb8
Merge pull request #118497 from mimowo/fix-kubelet-regression
Proceed with deletion of rejected pods after Kubelet restart
2023-06-23 14:05:07 -07:00
Marek Siarkowicz
22258a263f Fix stable metric finder for NewDesc with custom import name 2023-06-23 16:57:56 +02:00
Marek Siarkowicz
1384274622 Update metric documentation 2023-06-23 14:54:17 +02:00
Kensei Nakada
6f8d38406a feature(scheduler): implement ClusterEventWithHint to filter out useless events 2023-06-22 13:36:19 +00:00
Michal Wozniak
e3ee9b9adc Fix the deletion of rejected pods 2023-06-22 09:18:34 +02:00
Patrick Ohly
c91c578795 scheduler_perf: skip expensive cleanup during benchmarks
Each benchmark test case runs with a fresh etcd instance. Therefore it is not
necessary to delete objects after a run.

A future unit test might reuse etcd, therefore cleanup is optional.
2023-06-22 08:56:14 +02:00
Kubernetes Prow Robot
a532733703
Merge pull request #118754 from pacoxu/fix-prometheus-client
fix metrics test with 1.16.0 prometheus client
2023-06-21 20:25:39 -07:00
Kubernetes Prow Robot
3163606910
Merge pull request #118781 from tukwila/upgrade_addon_manager
upgrade image addon-manager/kube-addon-manager:v9.1.6 to v9.1.7
2023-06-21 16:49:52 -07:00
Kubernetes Prow Robot
28296ba59e
Merge pull request #113994 from mengjiao-liu/contextual-logging-controller-certificates
certificate controller: use contextual logging
2023-06-21 09:03:42 -07:00
Stanislav Laznicka
7f532891c9
e2e tests: set all PSa labels instead of just enforcing 2023-06-21 15:05:13 +02:00
Stanislav Laznicka
e5cbc51d29
e2e framework: allow setting all PSa labels at once 2023-06-21 13:02:01 +02:00
Patrick Ohly
ec70b2ec80 e2e dra: add "kubelet must skip NodePrepareResource if not used by any container"
If (for whatever reason) no container uses a claim, then there's no need to
prepare it.
2023-06-21 10:42:22 +02:00