Hey there! We've found the following might help you get unblocked faster:

• 🧵 mongodb container keeps crashing today, no load on the server (no access at all…)

• 🧵 Express App EADDRINUSE error with PM2 deployment

• 📚 Application Failed to Respond

If you find the answer from one of these, please let us know by solving the thread!

I'm looking at your service metrics and it appears like everything is within bounds. I will say though we do support hosting supabase on your railway project! Theres a template here:
https://railway.com/deploy/supabase

Are you able to try using that and seeing if that fixes it?

It just happened again 12 minutes ago same thing, just a random shutdown.

I printed the memory and it never went above 72%

I'll try the supabase thing tomorrow but I don't think it'll fix that, there's no way to investigate why those shutdowns are happening?

I don't see any erroneous logs on my end which would indicate a platform issue. This does feel a bit like an issue with how your app is exiting on a form of error. I'll open this to the community to help!

I really don't think my app is crashing... It's been working a long time when I was the only user and in local env. This really seems to be a load issue. It crashes all the time and on pretty minimal use. It doesn't make any sense. All signs point to a SIGTERM from the platform or from the DB. I also had this error at times:

psycopg2.OperationalError: SSL connection has been closed unexpectedly

And now it's down again... It can't continue like this, it just keeps happening. Again no error message. Working perfect and than boom shut down.

Bro the issue isn’t Railway the platform is killing your API because your Postgres connections are dropping and your API freezes. Do these two things and it will stop crashing

1. Add Postgres connection pooling - If you’re using Python, stop creating psycopg2 connections on every request.

use pool

from psycopg_pool import ConnectionPool

pool = ConnectionPool(conninfo=DATABASE_URL, min_size=1, max_size=20)

with pool.connection() as conn:

This prevents Supabase from randomly closing SSL connections.

2. Run your API with multiple workers. If you’re starting your server with uvicorn app:app than that’s only 1 worker, so one blocked DB call kills the whole app. Instead Start it like this instead - gunicorn app:app -k uvicorn.workers.UvicornWorker --workers 20 --timeout 60

   Multiple workers = no more random shutdowns when a DB connection drops.

Hey, let's go through this step by step as there could be many reasons why that's happening. Also, there's no need to use Supabase on Railway, so don't worry about that.

First of all, please share more logs about your application when it crashes. Do you see any pattern regarding when these crashes occur (any endpoints being called, traffic spikes, etc.)? Are you able to connect to your database after the crashes happen? Additionally, enabling your Supabase Postgres Logs is an effective way to identify who is terminating the connection: https://supabase.com/docs/guides/telemetry/logs#logging-postgres-queries

I would also recommend using sentry.io in your application as it's good to track these kinds of issues.