Skip to content

nodeagent DaemonSet delete missing errors.IsNotFound check, causes DeleteDaemonSetFailed / test flakiness #2372

Open

Description

@kaovilai

Bug

The node-agent DaemonSet deletion path doesn't check errors.IsNotFound(err) before treating a delete failure as a real error — unlike the adjacent ConfigMap deletion path in the same reconciler, which does handle this correctly.

// DaemonSet path (missing IsNotFound check):
if err := r.Delete(deleteContext, ds, &client.DeleteOptions{...}); err != nil {
    r.EventRecorder.Event(...)
    return false, err  // should treat IsNotFound as success (already gone)
}

Recommended fix: add errors.IsNotFound(err) handling after the r.Delete() call — when the delete fails with NotFound, treat it as success (the DaemonSet is already gone), matching the ConfigMap path's existing behavior.

Affected versions (confirmed present in all)

Branch File Line
oadp-dev internal/controller/nodeagent.go 278
oadp-1.6 internal/controller/nodeagent.go 278
oadp-1.5 internal/controller/nodeagent.go 214
oadp-1.4 controllers/nodeagent.go 152

How this was found

Surfaced by this repo's own automated e2e log-analysis tooling on ci/prow/4.23-e2e-test-aws for PR #2206 (oadp-1.6, unrelated change — veleroPredicate/annotation-change reconciliation): the AfterEach cleanup of "Configuration testing for DPA Custom Resource" (tests/e2e/dpa_deployment_suite_test.go) hit a DeleteDaemonSetFailed event while the velero pod was stuck PodInitializing, failing the test. The analysis explicitly concluded: "Likely Cause: Real bug in the OADP operator's DaemonSet reconciliation logic. Not a flake, not environmental." None of the known-flake patterns in tests/e2e/lib/flakes.go matched.

Log: https://prow.ci.openshift.org/view/gs/test-platform-results/pr-logs/pull/openshift_oadp-operator/2206/pull-ci-openshift-oadp-operator-oadp-1.6-4.23-e2e-test-aws/2087353890981810176

This appears to be a pre-existing, long-standing bug rather than something introduced by #2206 — it just happened to get exercised by that PR's e2e run.

Note

Responses generated with Claude

Activity

  1. openshift-ci commented on Nov 11, 2026

    @openshift-ci

    Issues go stale after 90d of inactivity.

    Mark the issue as fresh by commenting /remove-lifecycle stale.
    Stale issues rot after an additional 30d of inactivity and eventually close.
    Exclude this issue from closing by commenting /lifecycle frozen.

    If this issue is safe to close now please do so with /close.

    /lifecycle stale

  2. added
    lifecycle/staleDenotes an issue or PR has remained open with no activity and has become stale.
    on Nov 11, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    lifecycle/staleDenotes an issue or PR has remained open with no activity and has become stale.

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions