Hey, running into a weird issue with our staging service and hoping you can help figure out what's going on.

We have multiple replicas running and they're all connected to the same external MySQL database (verified the IP). Here's what's happening:

When we update the database directly, the API keeps returning old data mixed with new data. Like, the database has 1 row, but the API returns 2 rows - one old, one new.

The weird part is that restarting the Railway service immediately fixes it and everything returns the correct data. Our code doesn't have any caching implemented, so I'm wondering if there's something at the Railway infrastructure level that could be caching database connections or responses?

We've verified:

- All replicas are connecting to the same database IP

- The database itself has the correct data

- Our code doesn't cache anything

- Restarting fixes it instantly

Is there some kind of connection pooling or caching happening on Railway's side that could cause this? And if so, is there a way to configure it so replicas always pull fresh data without needing manual restarts?