🐞 Fatal error handling when using Diesel

[mwnDK1402]

When too many database connections are started, Diesel produces this error:

[ERROR r2d2] Too many connections

The problem is, that this error is being spammed hundreds of times, until the lambda function times out, after which it starts another function, which immediately gives the same error, looping again and again.

This has happened 15 times in 3 seconds, after which it stopped (somehow). The server stops functioning completely at this point, until I redeploy it.

The reason I'm submitting this here is that I have only ever seen this issue when running the application on Lambda, and the error handling is fatal.

[migeruj]

Hi, this also happens to me.
Rocket.rs is fully synchronous at the moment. When you use RDBMS on serverless the usual problem is sessions. When you use r2d2 to create a pool you are creating one connection per core. Check if your database runs out of connections. I think we should look for a way to handle a pool on serverless or wait for the rocket.rs 0.5 version (currently on dev) which is already set up for async RDBMS.

[mwnDK1402]

I'm already using Rocket 0.5. Still happens. I talked to the author of Rocket, and he said it was r2d2 handling the errors ungracefully.