Our production web service is unavailable after two deployments of the same build started normally and served successful requests, then changed from SUCCESS to FAILED about eight minutes later. No preceding application crash or out-of-memory error appeared in the inspected runtime logs. We have paused further retries.

Times on 31 October 2026, UTC:

• First deployment became FAILED at 14:36:35; its container was stopped at 14:36:49.
• One same-build recovery attempt was ready at 14:46:52, still served a successful health request at 14:54:26, and was stopped at 14:54:39.
• At 16:27, the public service still returned HTTP 502. The platform reported zero running replicas out of one and no pending operation.

Is there a known deployment-promotion or network-configuration issue matching this pattern, and a supported non-destructive recovery that preserves the persistent volume and data? We have not established the underlying cause.

This question is intentionally unlinked to a project or service to keep identifying details off the public thread. No domain, project/service/deployment IDs, raw logs, credentials, or customer data are included. Private technical threads are disabled on our Hobby plan. Please advise a supported confidential channel for project-specific details, or public platform-level guidance. Please do not make changes on our behalf from this unlinked report.