Custom recover at connection level

[bfwbbrj]

Custom recover at connection level

[bfwbbrj]

@FZambia

[FZambia]

Hello, I am not sure about this idea. Centrifuge itself must not panic - this should be fixed ASAP. I also have this opinion regarding running applications in production. Here with Centrifuge we have a more difficult situation then with HTTP requests - connections are stateful and in most cases we don't know what to do in case of panic. This may theoretically lead to unexpected side effects.

I suppose you can add recovery logic on your app level?

[bfwbbrj]

connections are stateful and in most cases we don't know what to do in case of panic. This may theoretically lead to unexpected side effects.

In case of panic, maybe we can just close the client that panics? This way we can ensure rest of the clients remain unaffected. Currently if this happens, all other connections are closed as well.

I suppose you can add recovery logic on your app level?

I did, but it didn't stop other clients from getting closed :/

[FZambia]

I did, but it didn't stop other clients from getting closed :/

Could you elaborate more on this? If you added recovery then it should fully solve the case - or am I missing sth here?