A single-copy deployment has two leaders
Weik.io 2026.2Two copies of a single-copy deployment led at the same time for 5 seconds or more. A file or a scheduled job may have been handled twice.
| Kind | Severity | Who sees it |
|---|---|---|
deployment- |
Error while both copies lead, then a warning | Everyone who can read the deployment |
What it means#
Two copies of a deployment that should run one copy at a time led it at the same time. Work that only the leader does may have run twice. For example, a file may have been picked up twice, or a scheduled job may have run on two servers.
What raises it#
The problem shows when two copies have led at the same time for 5 seconds or more. It’s an error while both copies still lead. Once the overlap ends, it becomes a warning whose detail has the overlap’s start and end. It leaves the problem list 15 minutes after the end.
A copy that lost contact with the hub and kept leading stands down when it reconnects.
What to check#
See the leader changes of the last seven days:
$ weikio describe deployment <project>/<integration> --env <environment>Read what the routes did while both led:
$ weikio logs <project>/<integration> --env <environment> --from <start> --to <end>How to fix it#
Check the work the routes did in the overlap, and undo or deduplicate what ran twice.
By default, a leader that can’t reach the hub keeps leading, but in a change-tracking integration it stands down. When doing the work twice is worse than pausing it, set whenHubUnreachable: standDown in integration.yaml. The leader then stands down 15 seconds after the hub last confirmed that it leads.