Stuck volume lock after scale-to-0-then-1 — need force-release of container process holding /data
bcervatti
PROOP

13 days ago

Production service down 40+ minutes. Scale-to-0-then-1 did NOT clear the stuck lock.

Service: preleilao (ID: b0c01512-ea5f-4dfc-a0db-ef3962e859ff)

Project: protective-luck (ID: ee10c4c6-3756-4871-a3d7-8d02a4f8908c)

Environment: production (ID: a5678bce-2ea3-4574-982c-9f71be2dcefb)

Region: us-west2

Volume mounted at: /data

Deployment timeline:

  • Original stuck deployment: 60f0ae60-ea95-401e-b602-4adb3d86d709 (stuck on CREATE_CONTAINER since 15:35:41Z)
  • Scale-down deployment: 836358ea-e9a2-4da1-8a39-efd0847e5329 (15:54:08Z)
  • Scale-up deployment: ae1d342d-ba9a-4375-8ad6-f82a4f8ef010 (15:54:12Z)

Current symptoms:

  • All requests return 502 with x-railway-fallback:true
  • Scale-to-0 completed but did NOT release volume lock
  • New containers still cannot attach to /data
  • Auto-deploy is disabled (no cascading redeploys)

Need: Server-side investigation and force-release of stuck container process holding /data volume lock. Not a redeploy retry — the lock is persisting even after complete shutdown.

This is NOT a code or config issue — the volume lock is physically held by infrastructure. Need forced cleanup.

Awaiting User Response

1 Replies

Status changed to Awaiting Railway Response Railway 13 days ago


chandrika
EMPLOYEE

12 days ago

Your read is right. A previous container did not shut down on our side, and it

is still holding the mount, which is why scaling back up did not release it and

why the replacement volume did not change the outcome. This is not caused by

anything in your project or your application code, so further redeploys will

not clear it. Both the original volume and the replacement you created on the

8th are healthy, and your data is intact on both. We are clearing the stuck

container from our side and will confirm here once you can redeploy.


Status changed to Awaiting User Response Railway 12 days ago


Welcome!

Sign in to your Railway account to join the conversation.

Loading...