https://bz.apache.org/bugzilla/show_bug.cgi?id=58624
--- Comment #7 from Barry Coughlan <b.coughl...@gmail.com> ---
(In reply to Mark Thomas from comment #6)
> Thanks for all the work on this.
>
> It is worth updating to the latest 8.0.x in case one of the I/O or
> concurrency updates has improved this.
>
> I'm switching this to NEEDINFO. Ideally we need a test case but that is not
> essential. If you believe you have found a code path that could trigger this
> let us know and we'll take a look. As long as we can confirm the code path
> is theoretically possible, we'll be happy to fix it.
Hi Mark,
I managed to replicate this on 8.0.24 and 8.0.28. I don't think I could
automate a test to replicate this, but it can be shown with a simple example by
setting breakpoints.
============================================================
Sample server endpoint:
------------------------
package tomcatbug58624;
import javax.websocket.EncodeException;
import javax.websocket.Encoder;
import javax.websocket.EndpointConfig;
public class WsEncoder implements Encoder.Text<Object> {
@Override
public void destroy() {
}
@Override
public void init(EndpointConfig endpointConfig) {
}
@Override
public String encode(Object object) throws EncodeException {
return (String) object;
}
}
------------------------
package tomcatbug58624;
import java.io.IOException;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import javax.websocket.CloseReason;
import javax.websocket.EncodeException;
import javax.websocket.OnClose;
import javax.websocket.OnError;
import javax.websocket.OnMessage;
import javax.websocket.OnOpen;
import javax.websocket.Session;
import javax.websocket.server.ServerEndpoint;
@ServerEndpoint(value = "/bug58624", encoders = WsEncoder.class)
public class WsEndpoint {
public static class SendWsMessage implements Runnable {
private Session session;
public SendWsMessage(Session session) {
this.session = session;
}
@Override
public void run() {
try {
session.getBasicRemote().sendObject("test");
} catch (IOException | EncodeException e) {
e.printStackTrace();
}
}
}
private static final ExecutorService ex = Executors.newFixedThreadPool(1);
@OnOpen
public void onOpen(Session session) throws IOException {
ex.submit(new SendWsMessage(session));
}
@OnMessage
public void onMessage(String message) {
System.out.println("OnMessage: " + message);
}
@OnError
public void onError(Throwable t) {
System.err.println("OnError:");
t.printStackTrace();
}
@OnClose
public void onClose(Session session, CloseReason cr) {
System.out.println("Closed " + cr);
}
}
-----------------------------
Set breakpoint on WsEndpoint$SendWsMessage:29 (The .sendObject() line)
Set breakpoint on WsRemoteEndpointImplServer:76 (start of doWrite())
- Open websocket
- Thread 1 breaks on sendObject()
- Close websocket from client side (I just did Ctrl+C with Tyrus CLI)
- Thread 2 breaks on doWrite()
- Continue Thread 1 so that the message gets queued to send
- Continue Thread 2
sendObject() waits indefinitely for Future.get(). The reason is that doWrite is
not set to true in endMessage() because closed = true.
This would happen any time a message is queued to be sent while a close is in
progress.
I’m not familiar enough with the codebase to submit a patch, but it looks to me
like if closed==true in endMessage(), it should call sendResult with an
exception for any messages still in the queue.
--
You are receiving this mail because:
You are the assignee for the bug.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org
