Skip to content

[CI] SLMStatDisruptionIT testCurrentlyRunningSnapshotNotRecordedAsFailure failing #155621

Description

@elasticsearchmachine

Build Scans:

Reproduction Line:

./gradlew ":x-pack:plugin:slm:internalClusterTest" --tests "org.elasticsearch.xpack.slm.SLMStatDisruptionIT.testCurrentlyRunningSnapshotNotRecordedAsFailure" -Dtests.seed=35A45A714ACDC0B -Dtests.locale=es-PH -Dtests.timezone=America/Ensenada -Druntime.java=26

Applicable branches:
main

Reproduces locally?:
N/A

Failure History:
See dashboard

Failure Message:

java.lang.AssertionError: expected:<0> but was:<1>

Issue Reasons:

  • [main] 1 consecutive failures in test testCurrentlyRunningSnapshotNotRecordedAsFailure
  • [main] 1 consecutive failures in step part2
  • [main] 1 consecutive failures in pipeline elasticsearch-intake

Note:
This issue was created using new test triage automation. Please report issues or feedback to es-core-infra.

👀

Activity

  1. elasticsearchmachine commented on Jul 31, 2026

    @elasticsearchmachine
    CollaboratorAuthor

    This has been muted on branch main

    Mute Reasons:

    • [main] 1 consecutive failures in test testCurrentlyRunningSnapshotNotRecordedAsFailure
    • [main] 1 consecutive failures in step part2
    • [main] 1 consecutive failures in pipeline elasticsearch-intake

    Build Scans:

  2. elasticsearchmachine commented on Jul 31, 2026

    @elasticsearchmachine
    CollaboratorAuthor

    Pinging @elastic/es-distributed (Team:Distributed)

  3. self-assigned this
    on Jul 31, 2026
  4. kingherc commented on Jul 31, 2026

    @kingherc
    Contributor

    Root cause: TOCTOU in SnapshotLifecycleTask.WriteJobStatus. When snapshot A completes, it looks up registered snapshots that are no longer running and fetches their SnapshotInfo. If snapshot B is still in SnapshotsInProgress at lookup time it is skipped; if B then finishes before A’s cluster-state update runs, B has no SnapshotInfo and is incorrectly counted as a failure (assertStats expected 0 failed, was 1). CI also shows the warning that B was removed from the registered set by another snapshot’s cleanup.

    This is a product bug (false SLM failure stats under concurrent completions), not a test-only flake. Marking as medium priority; not higher since it's for stats. Will follow up with reproduction and possible solution.

  5. added
    medium-riskAn open issue or test failure that is a medium risk to future releases
    on Jul 31, 2026
  6. added 2 commits that reference this issue on Aug 5, 2026
    ce069a2
    442e349
  7. added 3 commits that reference this issue on Aug 19, 2026
    8acd8ef
    6dde7c7
    bea291f
  8. added a commit that references this issue on Aug 20, 2026
    c71ee89
  9. added a commit that references this issue on Aug 24, 2026
    99d569d
  10. added a commit that references this issue on Sep 1, 2026
    e83faba
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

:Distributed/Snapshot/RestoreAnything directly related to the `_snapshot/*` APIs>test-failureTriaged test failures from CITeam:DistributedMeta label for distributed team.medium-riskAn open issue or test failure that is a medium risk to future releases

Type

No type

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions