Skip to content

Commit 6c8653c

Browse files
committed
test/e2e/upgrade: Raise default update-ack timeout to 10m
It's failing in [1]'s client-go bump to Kube 1.35, with timelines like [2]: * 11:17:27, test-suite patches ClusterVersion to request the update. * 11:17:27, CVO Events UpgradeStarted (although it hasn't actually pivoted the target yet, might want to adjust this Event name/timing) and RetrievePayload: Retrieving and verifying payload… * 11:17:28, CVO creates version–fgn9s. * 11:17:28, CVO starts watching version--fgn9s. * 11:17:33, version--fgn9s container started. So it's not slow image pulls. * 11:17:34, version--fgn9s container started again? Not clear what happened here. * 11:17:37, rename-to-final-location container started (the last container in version-... Pods, it should be an atomic, single-filesystem mv). * 11:17:38, rename-to-final-location exits 0, so success, but the CVO does not notice. * 11:19:27, test-case times out after 2m. * 11:21:27, CVO refreshes the watch on version--fgn9s. and CVO logs like: I0320 11:17:58.306414 1 reflector.go:1159] "Warning: event bookmark expired" err="k8s.io/client-go/tools/watch/informerwatcher.go:162: hasn't received required bookmark event marking the end of initial events stream, received last event 18.795159461s ago" I0320 11:18:08.306345 1 reflector.go:1159] "Warning: event bookmark expired" err="k8s.io/client-go/tools/watch/informerwatcher.go:162: hasn't received required bookmark event marking the end of initial events stream, received last event 28.795076029s ago" ... I0320 11:21:18.307818 1 reflector.go:1159] "Warning: event bookmark expired" err="k8s.io/client-go/tools/watch/informerwatcher.go:162: hasn't received required bookmark event marking the end of initial events stream, received last event 3m38.796537922s ago" I0320 11:21:27.596115 1 trace.go:236] Trace[23813503]: "Reflector WatchList" name:k8s.io/client-go/tools/watch/informerwatcher.go:162 (20-Mar-2026 11:17:28.301) (total time: 239294ms): 21 - 17 = 4m, and I'm giving it 10m to be safe. The bump allows work like [1] to move forward while we troubleshoot the client issue. We'll definitely want to revert this once the 1.35 clients are fixed. [1]: openshift/cluster-version-operator#1282 [2]: https://prow.ci.openshift.org/view/gs/test-platform-results/pr-logs/pull/openshift_cluster-version-operator/1282/pull-ci-openshift-cluster-version-operator-main-e2e-agnostic-ovn-upgrade-out-of-change/2034935978228977664
1 parent 1883631 commit 6c8653c

1 file changed

Lines changed: 1 addition & 1 deletion

File tree

test/e2e/upgrade/upgrade.go

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -82,7 +82,7 @@ var (
8282
// upgradeAbortAtRandom is a special value indicating the abort should happen at a random percentage
8383
// between (0,100].
8484
const upgradeAbortAtRandom = -1
85-
const defaultCVOUpdateAckTimeout = 2 * time.Minute
85+
const defaultCVOUpdateAckTimeout = 10 * time.Minute
8686

8787
// SetTests controls the list of tests to run during an upgrade. See AllTests for the supported
8888
// suite.

0 commit comments

Comments
 (0)