Logs full of "Unable to send auto-ping for WebSocket connection. WebSocket is closed"

[RobertoUa]

Describe the bug

We're using websockets-next with quarkus.websockets-next.server.auto-ping-interval=10s.
And sometimes it starts spamming logs with
Unable to send auto-ping for WebSocket connection [endpointId=..., path=/api/v1/..., id=...]: io.vertx.core.impl.NoStackTraceThrowable: WebSocket is closed

It seems that the timer sometimes is not canceled on connection close. It's even possible that it's even worse and no close callbacks are called or there's some connection leak.
I have logs on each type of callbacks, and I only see @OnOpen and then WebSocket is closed is right after that. No @OnError or @OnClose

Expected behavior

No response

Actual behavior

No response

How to Reproduce?

No response

Output of uname -a or ver

No response

Output of java -version

No response

Quarkus version or git rev

No response

Build tool (ie. output of mvnw --version or gradlew --version)

No response

Additional information

No response

[gsmet]

I see the timer being cancelled here:

https://github.com/quarkusio/quarkus/blob/main/extensions/websockets-next/runtime/src/main/java/io/quarkus/websockets/next/runtime/Endpoints.java#L248-L250

I'm not sure if the timing is right and if the timer should be cancelled before completing the close.
Could you check if this code is called?

I think it would probably help if you could prepare a reproducer (but I'm not sure it's easily doable).

Also I'm wondering if, when the issue is solved, we should demote these messages to debug.

@mkouba that's probably one for you.

[mkouba]

It seems that the timer sometimes is not canceled on connection close. It's even possible that it's even worse and no close callbacks are called or there's some connection leak.
I have logs on each type of callbacks, and I only see @OnOpen and then WebSocket is closed is right after that. No @OnError or @OnClose

Hm, an @OnClose callback should be definitely invoked when a connection is closed. Have you tried to enable the traffic logging to see if a connection is closed properly?

[RobertoUa]

But if it isn't closed properly, then what?
How do I make sure it is closed properly? (although everything could happen, it's network).

Also, it seems that any of the methods before vertx.cancelTimer(timerId); could throw/fail, making it run forever.

[mkouba]

But if it isn't closed properly, then what? How do I make sure it is closed properly? (although everything could happen, it's network).

Also, it seems that any of the methods before vertx.cancelTimer(timerId); could throw/fail, making it run forever.

I think that we need to find out more... and ideally be able to reproduce the problem with a simple app.

[RobertoUa]

Unfortunately, I have no idea how to reproduce it. It just randomly happens on prod. And not always, but when it does, it just spams forever.