Hi there,

How many connections do you presume to be making to the database? And has this increased recently?

If so, you can consider putting pgbouncer in front of your database.

Regards,

Christian

There's a new bug that appeared overnight. The PostGIS container decided to redeploy overnight and it failed. No manual action was taken, but it appears to be having trouble migrating the volume.
This is starting to cause a lot of bugs on one of the most critical services. I seriously wonder if Railway is a stable solution.

Heya, could you elaborate on issues this has caused ? while your migration failed, your database remains online and operational.

It's likely there was a blip during the migration failure which caused the error you sent.

Certainly my database is still operational but it is still worrying that deployments are happening without us even being at the initiative and that you cannot find it?

Hello Alex,

First off, our transition to Railway Metal has been well communicated ahead of time via our documentation pages and our email communications.

With that said, the ECONRESET errors that you are reporting are unlikely to do with Railway provided infrastructure. I see that you have two up redpployments for your DB and that would be likely to cause issues with your deployments.

There is not much else we can provide for you here as the DBs on Railway are not managed and could be affected by a variety of factors such as connections and load.