2 months ago
Yesterday my March Madness website experienced a problem getting corect information from the ESPN API, that required code revisions.
Antigravity made those code revisions correctly.
However, when we pushed the deployment using Railway, Railway overwrote a persistant database thereby destroying any possibility of recovery and proper deployment.
At minimum, if railway is going to overwrite a DB there must be a approval function before the DB is overwritten. That is fundamental function for all situations where data might be destroyed.
Because the data was overwritten the game had to be stopped and money had to be returned to the players.
Based on this unacceptable performance by Railway, I will never use this service again.
Has anyone else had this terrible experience with this half baked Railway product?
3 Replies
2 months ago
Deployments on our platform do not overwrite volume data. Volumes persist across deploys and are only affected if explicitly deleted or wiped. If your database was stored in a file (like SQLite) within the application directory but not on a mounted volume, a new deployment would replace it since the container is rebuilt fresh each deploy. To protect data going forward, ensure your database is stored on a volume and consider enabling scheduled backups for recovery options.
Status changed to Awaiting User Response Railway • 2 months ago
Railway
Deployments on our platform do not overwrite volume data. Volumes persist across deploys and are only affected if explicitly deleted or wiped. If your database was stored in a file (like SQLite) within the application directory but not on a mounted volume, a new deployment would replace it since the container is rebuilt fresh each deploy. To protect data going forward, ensure your database is stored on a [volume](https://docs.railway.com/volumes) and consider enabling [scheduled backups](https://docs.railway.com/volumes/backups) for recovery options.
2 months ago
We did that and it overwrote anyway
Status changed to Awaiting Railway Response Railway • 2 months ago
Status changed to Open Railway • 2 months ago
2 months ago
Hey, Railway volumes do persist across deploys, the issue is likely your app code. If your app creates/initializes the DB file on startup without checking if one already exists on the volume, it'll overwrite the persisted data. Make sure your code checks for the existing file before creating a new one.
Railway treats files on a volume all the same, so it can't differentiate between a DB file and any other file. The responsibility is on your app code to check if the file already exists.
Status changed to Solved medim • about 1 month ago