I was able to reproduce the failure rather easily although for some runs it took many thousands of iterations. I reviewed all the logging tests in test/java/util/logging and it appears that only the two that I fixed have the problem.
Jim ----- Original Message ----- From: mandy.ch...@oracle.com To: jim.g...@oracle.com Cc: core-libs-dev@openjdk.java.net Sent: Friday, October 12, 2012 6:59:45 PM GMT -05:00 US/Canada Eastern Subject: Re: Please review: 7146552 LoggingMXBeanTest intermittent failure On 10/12/2012 9:55 AM, Jim Gish wrote: > Please review > http://cr.openjdk.java.net/~jgish/Bug7146552-LoggingMXBeanTest-failure/ > <http://cr.openjdk.java.net/%7Ejgish/Bug7146552-LoggingMXBeanTest-failure/> > > > The LoggingMXBeanTest has been intermittently failing because garbage > collection can cause the (weak) references to Logger instances to go > away. LoggingMXBeanTest2 has the same issue, but no bug has (yet) > been filed against it. > > The fix is to simply make the Logger variables static so they don't > get gc'd. Looks okay to me. Like Alan said, you can use the same bug for both tests as it's the same issue. I checked the nightly failure statistics and these 2 tests have been passing for some time. Were you able to reproduce the intermittent failure (it may require to modify the test and force GC) and verify the fix? I am not up to whether all logging tests have been audit if they may run into the same issue per the suggestion [2] ([1] was the thread reviewing a similar test bug fix). It'd be good to look at all logging tests if you have the cycle. Thanks Mandy [1] http://mail.openjdk.java.net/pipermail/core-libs-dev/2011-November/008136.html [2] http://mail.openjdk.java.net/pipermail/core-libs-dev/2011-November/008148.html
