This error usually means Railway’s init process cannot exec the container’s real entrypoint. For a Postgres service on Railway, I would check two things before trying any data migration:

1. In the Postgres service, go to Settings -> Deploy and make sure there is no custom Start Command or Pre-deploy Command set. If anything like npm start, Prisma, Django, or app commands were accidentally applied to the Postgres service, clear those fields back to default.

2. Then redeploy the existing Postgres service, keeping the existing postgres-volume attached. Do not detach/delete the volume and do not create a replacement database for recovery yet.

This exact ERROR (catatonit:2): failed to exec pid1: No such file or directory has also been showing up during the May 19 Railway image-registry/deployment disruption. Railway’s current guidance on matching threads is to redeploy once deployments are working again, because redeploying pulls a fresh image. The data should remain on the attached persistent volume through the redeploy.

If it still crash-loops after that, share whether the Postgres service has any custom command configured and the image/tag shown for the service.

I've had the same today. I redeployed postgres instead of restarting it and it came back online

What happens if we dont have backups (like on Hobby plan)? Do we loose data when redeploying?

No, the data is stored on the volume.

Thank you so much! The fix worked perfectly. I went to Settings → Deploy, confirmed there was no custom Start Command set, then redeployed the Postgres service. It came back online and all my data is still intact. Really appreciate the fast and clear help!