On 05/07/2011 22:04, Mark Thomas wrote: > On 05/07/2011 21:38, Mark Thomas wrote: >> On 05/07/2011 21:33, ma...@apache.org wrote: >>> Author: markt >>> Date: Tue Jul 5 20:33:28 2011 >>> New Revision: 1143197 >>> >>> URL: http://svn.apache.org/viewvc?rev=1143197&view=rev >>> Log: >>> Not just this test failing. Need a better solution. Revert r1143126. >> >> There is something odd going on with these tests and my Linux test VM. >> It looks like something is going wrong at the JVM level. An update to >> 1.6.0_26 seems to have helped but hasn't fully fixed this. >> >> The common factor across all of the failures is that they go wrong in >> o.a.c.tribes.transport.ReceiverBase.bind in the recursive re-try calls >> (essentially all the retires fail even though there are ports available). >> >> I am going to experiment with different re-try approaches, probably >> introducing a little delay since it looks like a timing issue. >> >> I've tried and failed to produce a simple test case that recreates the >> bind failure. > > OK, with a silly amount of concurrency, I can reproduce this on Windows. > Next step, getting the test case to use ReceiverBase.
Or maybe not. Looks like I was hitting some sort of limit for open server sockets. Odd number though - 598. Anyway, I'm experimenting with some alternative ReceiverBase implementations. I may have something tomorrow. Mark --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
