a month ago
Project: desirable-peace (ID f2d479db-6e0d-4060-a6c3-9d1e2a13582b), service invoicer_chat_frontend, environment production, region EU West.
Every build of this service since ~July 11 10:00 UTC fails immediately. The build log contains only: scheduling build on Metal builder "builder-geecdm" — no code is fetched, no build steps run. This occurs identically for GitHub-triggered builds and railway up CLI uploads (verified today). Other services in the same project (invoicer_chat backend) build and deploy normally, including within the last hour.
Noted staff's recommendation [in a related thread] just now: canceled/redeployed via a fresh railway up upload. The latest build — deployment ID 06ae*..6ddb, service invoicer_chat_frontend, project f2d4..*582b — was scheduled onto the same node, builder-geecdm, and failed identically (log contains only the "scheduling build on Metal builder" line).
That's now 20+ consecutive builds over 3 days, all pinned to this one builder, across both GitHub-triggered and CLI-upload paths, while the backend service in the same project/region builds normally. Redeploying does not re-route us.
Please manually reassign this service's builds off builder-geecdm or pin us to another builder. The service is currently frozen on its July 11 image.
2 Replies
Status changed to Awaiting Railway Response Railway • about 1 month ago
a month ago
We can see that builds for invoicer_chat_frontend are not progressing past the scheduling stage in both production and staging, while your backend service is building normally.
As a potential workaround, try deleting the invoicer_chat_frontend service and recreating it with the same repo and settings. A fresh service may receive a different builder assignment. Since this service has no volume attached, there is no persistent data at risk.
Status changed to Awaiting Conductor Response sam-a • about 1 month ago
Status changed to Awaiting User Response sam-a • about 1 month ago
a month ago
Thanks for the quick reply, team!
The frontend is currently serving its July-11 image. Deleting the frontend service, loses that running deployment; so if the recreated service lands right back on builder-geecdm, the frontend would be fully down, not just stale - and I will have torn the service down for worse than nothing.
Before I do anything radical- please, do move the service off builder-geecdm (or exclude it for a recreate) and confirm back.
Appreciate your help!
Status changed to Awaiting Conductor Response Railway • about 1 month ago