2 months ago
Hi Railway team,
My PostgreSQL service hit a full-disk condition earlier today. I increased the volume size, and the Railway UI now shows the volume at around 50% usage, but the database still will not start.
The logs consistently show that PostgreSQL enters automatic recovery, reaches redo done, and then crashes with:
FATAL: could not write to file "pg_wal/xlogtemp.30": No space left on device
So from the UI it looks like there should be free space available, but inside the running PostgreSQL service it still behaves as if the volume is full.
Could you please check whether:
- the volume resize was actually applied to the PostgreSQL service instance that is trying to recover,
- the filesystem attached to the service reflects the new volume size correctly,
- a reattach / restart / platform-side action is needed to make the resized volume visible to PostgreSQL.
I do not currently have a usable backup, so I am trying to avoid wiping the volume if at all possible.
Thanks.
3 Replies
Status changed to Awaiting Railway Response Railway • about 2 months ago
2 months ago
Your volume resize from 500 MB to 1 GB didn't fully apply on our end, which is why PostgreSQL was still seeing the old, smaller disk size and crash-looping during WAL recovery.
We've corrected this on our side and redeployed your Postgres service. It should now have the full 1 GB of disk space available, and your data is intact.
Status changed to Awaiting User Response Railway • about 2 months ago
2 months ago
Sorry about the scare. There was a brief window where the volume resize didn't fully propagate to the filesystem, which caused the WAL recovery failures you saw. It's resolved now - your Postgres service is running and the volume is correctly sized at ~1 GB. We've shipped a fix to prevent this from happening on future resizes.
2 months ago
Thank you all! It's back ;)
Status changed to Awaiting Railway Response Railway • about 2 months ago
Status changed to Solved lorepanz • about 2 months ago