8 days ago
My Railway PostgreSQL service is offline, and the PITR recovery-time selector is unavailable. I need guidance on checking recoverability without overwriting or changing the original database volume.
What is the supported way to:
- Verify that the backup manifests and required WAL files form a complete, usable recovery chain while the source database is offline?
- Determine the latest recoverable timestamp when the PITR UI cannot select a recovery time?
- Restore into a separate, isolated PostgreSQL service and a new volume for validation, leaving the original volume and backups untouched?
Please explain any prerequisites and whether any step writes to the original volume. This is a request for guidance only, not authorization to restart, restore, delete, or modify any service, data, permissions, or billing. I am asking for the general safe procedure and am intentionally omitting project identifiers and customer data.
1 Reply
8 days ago
A Postgres point-in-time restore is the supported way to restore into a separate service, and it does not write to the original volume. You can run it from the CLI with railway postgres pitr restore --service <name> --at <timestamp> (optionally --new-service-name), so you don't need the dashboard's datetime picker.
The restore creates a new sibling Postgres service with its own new, empty volume. On first boot it pulls the most recent base backup at or before your target from the PITR archive bucket, replays the archived WAL up to that point, and promotes. It reads the source's archive but never archives into it, and the original service and volume are not touched.
On prerequisites: the restore window starts at the first base backup taken after PITR was enabled, and the service must be on an official Railway Postgres image with a major version tag. The new service also gets the archive bucket's credentials, which can write as well as read.
railway postgres pitr status shows restore coverage and archiver health. Its live coverage check connects to the running container over SSH, so it reports that check as unavailable while the source is offline.
We can't give you a separate offline check of the manifests and WAL chain, or the latest recoverable timestamp. The restored service is the practical test: if it reaches your target and starts, the chain was usable up to that point. Each attempt creates a new service and volume and never writes to the original.
Status changed to Awaiting User Response Railway • 8 days ago
18 hours 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 18 hours ago