Quarkus vertx issues with Vite dev mode connection

[MarcinVaadin]

Description of the bug

Discovered during Quarkus update:
2023-01-24 09:15:41,523 WARN [io.ver.cor.imp.BlockedThreadChecker] (vertx-blocked-thread-checker) Thread Thread[vert.x-eventloop-thread-3,5,main] has been blocked for 3895 ms, time limit is 2000 ms: io.vertx.core.VertxException: Thread blocked

Quarkus build works using java.net.preferIPv4Stack=true

On development mode connecting to dev server is not possible while using IPv6.

Related to #15689

Expected behavior

Localhost IP should not be hardcoded?

Minimal reproducible example

Build app in dev mode

Versions

• Vaadin / Flow version: 24.0-SNAPSHOT
• OS version: Any with IPv6

[Artur-]

Is IPv4 not available? Vite is configured to bind to 127.0.0.1

    server: {
      host: '127.0.0.1',

but based on the logs it seems like it does not bind to that but to IPv6?

[Artur-]

It would probably be possible to parse the

Local: http://host:port/VAADIN/

from the Vite output and always use the address from there, then it would work no matter how it is bound

[MarcinVaadin]

Broader stacktrace:
2023-01-24 11:55:02,088 WARN [io.ver.cor.imp.BlockedThreadChecker] (vertx-blocked-thread-checker) Thread Thread[vert.x-eventloop-thread-0,5,main] has been blocked for 604084 ms, time limit is 2000 ms: io.vertx.core.VertxException: Thread blocked at java.base@17.0.2/jdk.internal.misc.Unsafe.park(Native Method) at java.base@17.0.2/java.util.concurrent.locks.LockSupport.park(LockSupport.java:211) at java.base@17.0.2/java.util.concurrent.CompletableFuture$Signaller.block(CompletableFuture.java:1864) at java.base@17.0.2/java.util.concurrent.ForkJoinPool.unmanagedBlock(ForkJoinPool.java:3463) at java.base@17.0.2/java.util.concurrent.ForkJoinPool.managedBlock(ForkJoinPool.java:3434) at java.base@17.0.2/java.util.concurrent.CompletableFuture.waitingGet(CompletableFuture.java:1898) at java.base@17.0.2/java.util.concurrent.CompletableFuture.get(CompletableFuture.java:2072) at com.vaadin.base.devserver.viteproxy.ViteWebsocketConnection.<init>(ViteWebsocketConnection.java:73) <---- HERE at com.vaadin.base.devserver.viteproxy.ViteWebsocketProxy.<init>(ViteWebsocketProxy.java:55) at com.vaadin.base.devserver.viteproxy.ViteWebsocketEndpoint.onOpen(ViteWebsocketEndpoint.java:95) at io.undertow.websockets.EndpointSessionHandler.connected(EndpointSessionHandler.java:119)

[mcollovati]

It seems like a deadlock in the onMessage callback provided by ViteWebsocketProxy to ViteWebsocketConnection when there are many concurrent clients, such as in Quarkus tests

The following one blocks on sendText

browserSession.getBasicRemote().sendText(msg);
getLogger().debug("Message sent to browser: " + msg);

Refactoring it with asynchronous call seems to work and Quarkus test pass

browserSession.getAsyncRemote().sendText(msg, result -> {
    getLogger().debug("Message sent to browser: " + msg);
});

Just to clarify, the deadlock happens on Undertow stack

[Artur-]

If that helps, I see no reason why not to change it

[mcollovati]

The problem with IPv6 is probably because the development mode integration tests modules set -Dvaadin.devmode.vite.options=--host.
This configuration is not needed anymore, not that we proxy the Vite websocket

[vaadin-bot]

This ticket/PR has been released with Vaadin 24.0.0.alpha10 and is also targeting the upcoming stable 24.0.0 version.