a month ago
Hello Railway team,
Our PostgreSQL service fails every deployment during the Initialization stage and never reaches Deploy or Post-deploy. The dashboard only shows "Deployment failed", with no message and no logs. We know about today's threads on volume attachment (a change on your side that was rolled back); our symptom looks different and our third attempt, made after the rollback, failed the same way. Please investigate on your side and tell us how to recover the service while fully preserving the volume and its data.
IDS (non-secret)
- Project: trustworthy-radiance — aa735449-36f1-4413-8e83-e4a64cbe6768
- Environment: production — 7071c417-fe02-41d7-b9cc-719d84dd998a
- Service: Postgres — 022ef2b5-123f-4dcd-97b5-57c739b844bf (EU West / Amsterdam)
- Image: ghcr.io/railwayapp-templates/postgres-ssl:18 (sha shown in the panel: 469c7)
- Volume: postgres-volume — 256cfcfb-d824-46e0-afee-127a15e85806 (mount /var/lib/postgresql/data)
WHAT THE DASHBOARD SHOWS
- Initialization fails at 00:00. Deploy and Post-deploy remain "Not started".
- Message: "Deployment failed", nothing else. Deploy logs: "No logs in this time range".
- The "Diagnose" button ran twice (narrow and wide viewport) and returned no information.
THREE FAILED DEPLOYMENTS (GMT-3 / UTC)
- 2026-10-06 09:30:28 / 12:30:28 — 6555284c-59e2-4cc7-8f0a-d058b126dc58 (Activity: "Redeployment failed" at 09:30:28.676)
- 2026-10-06 10:32:56 / 13:32:56 — bc5f4b02-577c-4c54-ab5d-afc09ac614e6 (we used "Redeploy deployment", "exact same configuration")
- 2026-10-06 22:35:02 / 2026-10-07 01:35:02 — 0167de64-fdb0-4c74-85af-969052f17b87 (same "Redeploy deployment"), made AFTER the rollback below
ROLLBACK
- In https://station.railway.com/questions/still-failing-after-the-rollback-pre-de-b6b65f5b a Railway employee wrote that the change had been rolled back; the thread shows 2026-10-06 21:14 GMT-3 (2026-10-07 00:14 UTC). A user in that thread also cites an earlier rollback announcement at 20:47 UTC.
- status.railway.com showed no active incident shortly before our third attempt. Our third attempt (01:35 UTC) is after both times.
LAST HEALTHY STATE
- Last deployment that ran: 896871b0-8b2c-4341-a340-3528ee62c7ed (from 2026-09-09 00:01 GMT-3).
- Its log ends with a clean shutdown on 2026-10-04 17:06 GMT-3 (20:06 UTC): "sending signal SIGTERM to container" -> "received smart shutdown request" -> "checkpoint complete" -> "database system is shut down".
- We do not see an Activity event that explains that SIGTERM.
TIMELINE ON 2026-10-06
- 09:30:28.676 Postgres "Redeployment failed"; 09:30:28.795 ERP-Bofedal (second service in the project) "Redeployment failed".
- 09:30:29.706 Railway applied "Set services to serverless as part of free plan limits" (Sleep Application = true) to both Postgres and ERP-Bofedal. The change record says "Postgres, ERP-Bofedal deployed", so we cannot tell whether it is a cause or a consequence.
CONTEXT
- ERP-Bofedal (GitHub source + Dockerfile, no volume) in the same project DID build and start at 09:58 GMT-3 the same day, with Serverless enabled.
- postgres-volume is still attached to Postgres (the panel says "Cannot delete volume required by connected service").
- We have NOT run wipe, delete, recreate, detach/attach or re-initialize on the volume. We have NOT changed the image, any variable, PGDATA, Serverless, the plan or any PostgreSQL setting. The only extra operations were the two "Redeploy deployment" above.
- Activity shows no volume, image or resource changes for this service. Its only earlier changes were to one password variable (2026-09-08 and 2026-09-09).
- The Backups tab says backups require the Pro plan, so we have none.
WHAT WE ASK YOU TO DETERMINE
- Why does Initialization fail without a message or visible logs?
- Is there an internal orchestration/deployment error that the dashboard does not show?
- Is it related to the volume, the automatic switch to Serverless, the PostgreSQL image, or a recent platform change?
- Is the volume correctly attached, accessible and intact?
- What is the safe procedure to recover PostgreSQL while fully preserving the volume and its data?
Please do NOT perform any destructive operation on the volume (wipe, delete, recreate, re-initialize, detach/attach, or equivalent) without our explicit written authorization.
This message contains no credentials, connection strings, variable values or data. Tell us if you need any other non-secret identifier.
Thank you.
1 Reply
Status changed to Awaiting Railway Response Railway • about 1 month ago
a month ago
This isn't related to the volume attachment issue, and your volume is fine: postgres-volume is attached to Postgres and ready, with about 122 MB of 500 MB used. Your workspace is on the Free plan, and Free plan deployments have to run with Serverless enabled. "Redeploy deployment" reuses the exact configuration saved with the old September 9 deployment, which had Serverless off, so each redeploy is refused before a container starts. That's why there are no logs. Serverless is now enabled on the service, so start a fresh deploy that uses the current settings: press Ctrl+K (Cmd+K on Mac) and choose "Redeploy source image". Nothing in this touches the volume.
Status changed to Awaiting User Response Railway • about 1 month ago
a month ago
This thread has been marked as solved automatically due to a lack of recent activity. Please re-open this thread or create a new one if you require further assistance. Thank you!
Status changed to Solved Railway • about 1 month ago