Memory leak? after unclean connection shutdown about pgcat

[drdrsh]

Maybe because we are not calling release in this case
?

[levkk]

Very good find, fix incoming.

[levkk]

Fix merged. Let me know if it re-occurs or something else pops up.

[drdrsh]

Thanks @levkk

One more question though, I see there are some codepaths where we return early (e.g. https://github.com/levkk/pgcat/blob/main/src/client.rs#L366) , do we also want to release there?

Maybe make the call to release as a part of the Drop implementation for the client.

[levkk]

Good question @drdrsh ! Take a look here, we release on error. The corner case was not releasing on early successful return. Nonetheless, I agree with you, this would be better if we release in the Drop implementation. I'll make that change.

Thanks!

P.S. Thinking some more, we also release when we finish a transaction yet the Client object is still used, so maybe releasing in Drop may not be optimal. That being said I think the leak should be plugged now :)