Bug Report: Unable to write GOAWAY

[mkarg]

Brief Summary

If a HTTP/2 client process is forcably killed (like pressing CTRL-C in Windows Terminal), Payara prints a stack trace into server.log.

This stack trace makes not much sense, as this is not a bug, it is expected behavior in HTTP/2. Neither the server administrator nor the server developer can't do anything to prevent clients from behaving like this. So while it might be interesting to see this event in the log, there is no benefit from having a stack trace.

Expected Outcome

Either no log entry at all, or just a warning like Client closed TCP connection before server could send GOAWAY..

Current Outcome

[2025-03-26T09:50:00.544+0100] [Payara 6.2025.1] [WARNUNG] [] [org.glassfish.grizzly.http2.Http2Session] [tid: _ThreadID=85 _ThreadName=http-thread-pool::http-listener-1(2)] [timeMillis: 1742979000544] [levelValue: 900] [[
  Unable to write GOAWAY.  Terminating session.
java.io.EOFException
        at org.glassfish.grizzly.nio.transport.TCPNIOTransport.read(TCPNIOTransport.java:594)
        at org.glassfish.grizzly.nio.transport.TCPNIOTransportFilter.handleRead(TCPNIOTransportFilter.java:59)
        at org.glassfish.grizzly.filterchain.TransportFilter.handleRead(TransportFilter.java:133)
        at org.glassfish.grizzly.ssl.SSLBaseFilter$SSLTransportFilterWrapper.handleRead(SSLBaseFilter.java:927)
        at org.glassfish.grizzly.filterchain.ExecutorResolver$9.execute(ExecutorResolver.java:88)
        at org.glassfish.grizzly.filterchain.DefaultFilterChain.executeFilter(DefaultFilterChain.java:246)
        at org.glassfish.grizzly.filterchain.DefaultFilterChain.executeChainPart(DefaultFilterChain.java:178)
        at org.glassfish.grizzly.filterchain.DefaultFilterChain.execute(DefaultFilterChain.java:118)
        at org.glassfish.grizzly.filterchain.DefaultFilterChain.process(DefaultFilterChain.java:96)
        at org.glassfish.grizzly.ProcessorExecutor.execute(ProcessorExecutor.java:51)
        at org.glassfish.grizzly.nio.transport.TCPNIOTransport.fireIOEvent(TCPNIOTransport.java:510)
        at org.glassfish.grizzly.strategies.AbstractIOStrategy.fireIOEvent(AbstractIOStrategy.java:82)
        at org.glassfish.grizzly.strategies.WorkerThreadIOStrategy.run0(WorkerThreadIOStrategy.java:83)
        at org.glassfish.grizzly.strategies.WorkerThreadIOStrategy$WorkerThreadRunnable.run(WorkerThreadIOStrategy.java:101)
        at org.glassfish.grizzly.threadpool.AbstractThreadPool$Worker.doWork(AbstractThreadPool.java:535)
        at org.glassfish.grizzly.threadpool.AbstractThreadPool$Worker.run(AbstractThreadPool.java:515)
        at java.base/java.lang.Thread.run(Unknown Source)
]]

Reproducer

@GET @Path("helloworld")
public String helloworld() {
    return "helloworld";
}

On Windows install original cURL (don't use curl Powershell alias), start cURL --http2 https://my-payara.my-domain.local/helloworld?i=[1-100000] then press CTRL-C to hard-kill that task.

Operating System

Windows 10 Pro

JDK Version

Zulu 17 (64 Bit)

Payara Distribution

Payara Server Full Profile

[skipper4]

Hello @mkarg,

I was able to reproduce the issue and I have escalated this to the platform development team as Jira task FISH-10889.

Thank you for the report.

[mkarg]

Any news?

[mkarg]

As an Open Source Contributor I do understand very well that everybody has a lot of work, but would it be possible to keep up minimum social customs and at least reply with some brief info like "problem identified, solution in progress" instead of being totally silent for another week? To work together on Payara, we need to stay mutually informed. Thanks! 🤔

[mkarg]

Another week without any response. 🙁 Is it really too much to ask for to at least tell that the solution needs another month or two? 🤔

[mkarg]

@smillidge Steve, can you please ask your staff to keep their open source peers at least minimally informed? We all are very busy, but writing just "planned for 2025-12" would be enough and costs not more than five seconds. Thanks. (Markus, Jersey Contributor / Jakarta REST Committer).