[Bug]: Failed deployment never stops - no new deployment possible

[bigbasti]

Existing issue

• I have searched the existing issues and found no duplicate.

Problem description

One of the deployments failed for some reason - cant find any logs or any message why.

But the bigger problem is that the failed deployment is in "Stopping" state for more than 10 days now..

There is no way to get rid of it and no more deployments are possible. Only way is to remove the whole project and re create it.

Steps to reproduce

I don't know why the deployment failed as i cant find any logs.
For some reason deployment fails to stop.

Environment

• OpenShip version/commit: 0.7.2
• Deployment method: docker
• Operating system: ubuntu
• Browser (if applicable): Edge

Logs and screenshots

Additional context

No response

[Hydralerne]

I reproduced both a worker-completion defect and build-counter defects on current main. The corrections are in #940, awaiting merge.

• A failed database acknowledgement could discard the finished worker while leaving its execution lease open. The fix retries that acknowledgement without repeating deployment or cleanup. A worker that is still touching the host keeps its lease.
• Cancellation now records its status and duration together. Late worker results cannot reset that duration or undo cancellation, and terminal status reads no longer trust a stale in-memory phase over the recorded outcome.
• The dashboard now uses one clock per deployment attempt. Completed timers stay stopped across refreshes, retries do not inherit earlier attempts, and missing historical measurements show as unknown instead of elapsed days. A completed cancellation no longer becomes “Stopping” again when reopened.

Validation: 9,476 tests across the API, database, platform, and dashboard pass locally, with all four TypeScript checks passing. The regressions include real PGlite transactions and mounted React views. CI for the updated commit is here: https://github.com/oblien/openship/actions/runs/36002581884.

The screenshot alone still does not establish why your original worker stayed stuck for ten days, so I am keeping this report open. Please share:

• The affected deployment ID and the exact error or disabled state shown when trying to deploy again.
• A short excerpt from the Openship API container logs around the original failure/cancellation, especially completion-acknowledgement, fatal, cancellation, or connection errors.
• Whether the API has restarted since the deployment became stuck, whether “Stopping” persists after restart, and the current version/image tag if you upgraded from v0.7.2.

Please redact credentials and environment values. These details, or a fresh verification after deploying #940, will establish whether the original stall is covered or another cancellation path needs attention.