On 20/03/2015 18:47, Mark Thomas wrote: > On 18/03/2015 22:38, Rainer Jung wrote: >> Am 18.03.2015 um 22:50 schrieb Mark Thomas: >>> On 18/03/2015 13:52, Rainer Jung wrote: >>>> Just a heads up: Trunk, TC 8 and TC 7 fail on Gump for various >>>> connectors sporadically at >>>> >>>> Testcase: testBug57621 took 3.145 sec >>>> FAILED >>>> null >>>> junit.framework.AssertionFailedError >>>> at >>>> org.apache.coyote.http11.TestAbstractHttp11Processor.testBug57621(TestAbstractHttp11Processor.java:796) >>>> >>> >>> There are various tests that appear to have issues at the moment. >>> >>> Now I've cleared the Bugzilla backlog ready for the next 8.0.x release I >>> plan to spend some time on the CI systems. >> >> Thanks. >> >>> One of the things I want to do is move the BuildBot builds to a less >>> loaded slave. That should result in reduced run times for the builds. >> >> That specific test also fails for me locally. My test system is not a >> fast one but also not really loaded. > > I'm making some progress by running the tests on vmgump. The sequence of > events is: > - client sends first part of request > - server processes it > - client sends end of first request and start of second > - server processes end of first request > (start of second request still in read buffer) > - something goes wrong and read buffer position is reset to zero > - client sends end of second request > - server processes end of second request as request line and rejects it > > I haven't yet figured out why it goes wrong but I'm hopeful that I'll > get to the bottom of this today.
Fixed. See r1668135. It was a bug with pipe-lined request handling and the completion of async requests. Mark --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
