Issue Summary

Pipeline Cache reupload attempt on failure does not work due to lock

Build will show as successful when cache upload failed.

Steps to Reproduce

1. Run a pipeline to upload cache
2. Stop the pipeline whilst the cache is being uploaded to media services.

Expected Result

• Cache gets uploaded successfully in the retry

• Subsequent Pipelines shows Upload failed in Build Teardown step.

Actual Result

• Before the cache begins upload, we acquire a lock (to prevent other steps in parallel uploading the same cache and causing a conflict). This lock is held whilst the upload happens, then released. If a step is stopped, we are not correctly releasing the lock, and when we retry on failure, retries fail due to the lock being held. After a period of time, this lock is eventually released.

Workaround

There's no known direct workarounds. You'll have to wait approximately 30minutes to an hour for the next build to upload the cache successfully again.

If the issue persists even after an hour please feel free to raise a ticket with us: https://support.atlassian.com/contact/#/