2014-06-10 18:06 GMT+04:00 Mark Thomas <ma...@apache.org>: > On 10/06/2014 10:52, ma...@apache.org wrote: >> Author: markt >> Date: Tue Jun 10 09:52:56 2014 >> New Revision: 1601590 >> >> URL: http://svn.apache.org/r1601590 >> Log: >> Pull up AsyncTimeout >> APR/native used a separate flag to stop the thread. This avoided issues >> where running was set true->false-true in quick succession. This fix is now >> available to NIO2 and BIO as well. >> Note NIO does not (currently) use the AsyncTimeout > > The Async unit test is currently taking 15 mins per test. I'm guessing > that I broke something in the async timeouts with this refactoring. I'm > looking now.
http://svn.apache.org/r1601443 is suspicious for me. Was there a reason to add "if (socket.isAsync())" around timeout processing? The subsequent Gump run this morning aborted the BIO test for running longer than 60min. http://vmgump.apache.org/gump/public/tomcat-trunk/tomcat-trunk-test-bio/index.html (I looked at catalina.log file of that run and there were intervals >10 minutes between some starts and stops, but I do not know which tests are those). E.g. two subsequent lines: 10-Jun-2014 15:23:29.562 INFO [main] org.apache.coyote.AbstractProtocol.start Starting ProtocolHandler ["http-bio-127.0.0.1-auto-17-38746"] 10-Jun-2014 15:40:14.677 INFO [main] org.apache.coyote.AbstractProtocol.pause Pausing ProtocolHandler ["http-bio-127.0.0.1-auto-17-38746"] Best regards, Konstantin Kolinko --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
