Open upodroid opened 8 months ago
Have we stopped running presubmits? I stopped seeing them in Github PRs and https://testgrid.k8s.io/sig-etcd-presubmits seems empty. Nothing in https://prow.k8s.io/?repo=etcd-io%2Fetcd too
There is a thread in #testing-ops on Slack to investigate this issue https://kubernetes.slack.com/archives/C7J9RP96G/p1700688510160169
Interesting flake in unit test:
=== FAIL: storage/schema TestMigrate/Upgrading_3.6_to_v3.7_is_not_supported (0.01s)
logger.go:130: 2023-11-23T15:59:05.436Z WARN failed to preallocate an initial WAL file {"path": "/tmp/TestMigrateUpgrading_3.6_to_v3.7_is_not_supported2935238209/002/etcd_wal_test401442[383](https://prow.k8s.io/view/gs/kubernetes-jenkins/pr-logs/pull/etcd-io_etcd/17008/pull-etcd-unit-test/1727717754938593280#1:build-log.txt%3A383)7/wal.tmp/0000000000000000-0000000000000000.wal", "segment-bytes": 64000000, "error": "no space left on device"}
schema_test.go:207: Failed to create WAL: no space left on device
--- FAIL: TestMigrate/Upgrading_3.6_to_v3.7_is_not_supported (0.01s)
Does that unit test expect the pod to have an ephemeral volume of a specific size?
I think problem might stem from etcd WAL tests. I don't think that unit tests mock storage, just write to t.TempDir()
(should be /tmp/
by default). WAL creation pre-allocates 64MB, so if there are couple of such tests running without cleanup we could be allocating couple of hundreds of megabytes.
cc @siyuanfoundation
we could mount an emptyDir (disk or memory) to /tmp if etcd tests are writing to it heavily.
The Kubernetes project currently lacks enough contributors to adequately respond to all issues.
This bot triages un-triaged issues according to the following rules:
lifecycle/stale
is appliedlifecycle/stale
was applied, lifecycle/rotten
is appliedlifecycle/rotten
was applied, the issue is closedYou can:
/remove-lifecycle stale
/close
Please send feedback to sig-contributor-experience at kubernetes/community.
/lifecycle stale
/lifecycle frozen
Note for robustness tests, there is one functionally to github actions that was not migrated. Uploading of test report artifacts.
etcd DNS was migrated in https://github.com/kubernetes/k8s.io/issues/6600.
Today I have migrated the etcd netlify site to the Kubernetes account, and requested the CNCF close the etcd netlify account in CNCFSD-2245
Hello 👋 I am willing to take up some of the workflow issues for migration to Prowjobs
etcd is now a subproject of Kubernetes and etcd maintainers are looking to adopt the CI system and Infra management approach we use for Kubernetes.
If I missed something, feel free to comment on the issue and I'll update the tracker.
/cc @jmhbnz @serathius @wenjiaswe @mrbobbytables @ahrtr @ameukam @BenTheElder
/sig etcd /sig testing /priority important-soon /kind feature