13 hours ago
My PostgreSQL proxy at shuttle.proxy.rlwy.net:19062 has been returning
ECONNRESET on every connection since ~21:44 UTC May 20. Project:
fearless-bravery, environment: production.
Error: read ECONNRESET at TCP.onStreamRead
I already migrated my app to another provider but need a data dump from this
specific database to recover historical production data. The proxy was working
fine before that time.
Can Railway provide access to the data or a pg_dump of this database?
4 Replies
Status changed to Awaiting Railway Response Railway • about 13 hours ago
10 hours ago
Your xoc-db service appears healthy but its TCP proxy has likely become stale. Redeploying the database service from the three-dot menu on the service card should restore proxy connectivity. Once you can connect again, you can run pg_dump yourself to export the data. We don't provide data dumps directly, as databases on the platform are unmanaged.
Status changed to Awaiting User Response Railway • about 10 hours ago
Status changed to Solved mykal • about 10 hours ago
mykal
Your xoc-db service appears healthy but its TCP proxy has likely become stale. Redeploying the database service from the three-dot menu on the service card should restore proxy connectivity. Once you can connect again, you can run pg_dump yourself to export the data. We don't provide data dumps directly, as databases on the platform are unmanaged.
2 hours ago
Redeploying xoc-db fixed connectivity to shortline.proxy.rlwy.net:19101 but that database is empty. My production data is in a different database at shuttle.proxy.rlwy.net:19062 (password: qkCJLiyfhqnEMCKwIZQiLdEomMBcPCrY) which still returns ECONNRESET. This database does not appear as a service card in my Railway dashboard — it may be a legacy managed database. Can you help restore access to that specific proxy/database?
Status changed to Awaiting Railway Response Railway • about 2 hours ago
elbartox
Redeploying xoc-db fixed connectivity to shortline.proxy.rlwy.net:19101 but that database is empty. My production data is in a different database at shuttle.proxy.rlwy.net:19062 (password: qkCJLiyfhqnEMCKwIZQiLdEomMBcPCrY) which still returns ECONNRESET. This database does not appear as a service card in my Railway dashboard — it may be a legacy managed database. Can you help restore access to that specific proxy/database?
2 hours ago
The redeploy found 433 MiB of data in the volume and PostgreSQL was ready to accept connections. But then it stopped immediately ('Stopping Container') and restarted — on the second start the database was empty (7.5 MB). The logs show: pgbackrest: RESTORED_MARKER=missing which may have triggered re-initialization. Can you recover the volume data (vol_vyrl87hk6kjthz97) from before the restart? Also, can you tell me what happened to the database at shuttle.proxy.rlwy.net:19062? That was a different database (different password) that still returns ECONNRESET.
an hour ago
This appears to be related to your GCP outage and the 'builder storage fix' deployed at 21:28-21:35 UTC May 20. The timing matches exactly — our database proxy started failing at ~21:44 UTC, right after your fix was deployed. The PostgreSQL volume (vol_vyrl87hk6kjthz97) found 433 MiB of data on first restart but was re-initialized empty on the second restart due to RESTORED_MARKER=missing. Since this data loss was caused by your infrastructure incident, please restore the volume from your internal backups