On 11/11/2020 09:11, build...@apache.org wrote: > The Buildbot has detected a new failure on builder tomcat-trunk while > building tomcat. Full details are available at: > https://ci.apache.org/builders/tomcat-trunk/builds/5549 > > Buildbot URL: https://ci.apache.org/ > > Buildslave for this Build: asf946_ubuntu > > Build Reason: The AnyBranchScheduler scheduler named 'on-tomcat-commit' > triggered this build > Build Source Stamp: [branch master] 20edb410286fdfe141dc0178f361493d4f3ca72c > Blamelist: Martin Tzvetanov Grigorov <mgrigo...@apache.org> > > BUILD FAILED: failed compile_1
This is the TestWebSocketFrameClient.testConnectToRootEndpoint test again. I've made a little progress with this over the last day or so. What I know so far: - The first WebSocket request is processed - The second is sent to the server (on a new connection) - The acceptor on the server accepts the new connection - The new connection is added to the Selector and a key issued - A short time later the Selector has no keys - The connection is never selected with a valid attachment This suggests a something closing the socket or (less likely) a Selector bug. There appears to be a timing aspect to the failure. Adding and removing debug logging is impacting my ability to recreate the failure. I am able to trigger the issue ~20% of the time on a full NIO test run. I am currently working on trying to recreate the issue more reliably and on shorter test runs. At the moment, it feels like I am making some progress so I plan to keep working on this rather than tagging 10.0.x etc. I don't want to delay the tags too long so my current thinking is keep working on this bug for today and tag tomorrow. Any (lack of) progress may change that view. Mark --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org