Re: [JENKINS] Lucene-Solr-SmokeRelease-5.x - Build # 316 - Failure

2015-09-22 Thread Michael McCandless
Thank you for finding it and fixing it Dawid! Mike McCandless http://blog.mikemccandless.com On Tue, Sep 22, 2015 at 7:03 AM, Dawid Weiss wrote: > This was a bug, Mike. Thanks for being so diligent and noticing this. > > Dawid > > On Thu, Sep 10, 2015 at 12:26 AM,

Re: [JENKINS] Lucene-Solr-SmokeRelease-5.x - Build # 316 - Failure

2015-09-22 Thread Dawid Weiss
This was a bug, Mike. Thanks for being so diligent and noticing this. Dawid On Thu, Sep 10, 2015 at 12:26 AM, Michael McCandless wrote: > OK thanks for the analysis Dawid. > > Mike McCandless > > http://blog.mikemccandless.com > > > On Wed, Sep 9, 2015 at 5:58 PM,

Re: [JENKINS] Lucene-Solr-SmokeRelease-5.x - Build # 316 - Failure

2015-09-09 Thread Dawid Weiss
This does look suspicious, Mike. Looks like interrupt() on testLiveMaxMergeCount somehow escaped and the next test started... I'll have to take a look at the code to see how (and if) this is possible. It shouldn't be. D. On Wed, Sep 9, 2015 at 12:02 PM, Michael McCandless

Re: [JENKINS] Lucene-Solr-SmokeRelease-5.x - Build # 316 - Failure

2015-09-09 Thread Michael McCandless
Spooky failure in TestConcurrentMergeScheduler.testLiveMaxMergeCount: it seems like IW.forceMerge failed to notice that in fact all merges had completed, so the calling thread waited forever for it to finish... But, in the "interrupting leaked threads", why is there a "still running" thread from

Re: [JENKINS] Lucene-Solr-SmokeRelease-5.x - Build # 316 - Failure

2015-09-09 Thread Dawid Weiss
Hi Mike. I know what caused it. It's a "test suite" timeout, but the runner still attempts to execute any remaining tests, even after the timeout watchdog killed the test that hung. It's a bug in the runner in other words (but it doesn't change the fact that the original test did hang). Dawid

Re: [JENKINS] Lucene-Solr-SmokeRelease-5.x - Build # 316 - Failure

2015-09-09 Thread Michael McCandless
OK thanks for the analysis Dawid. Mike McCandless http://blog.mikemccandless.com On Wed, Sep 9, 2015 at 5:58 PM, Dawid Weiss wrote: > Hi Mike. > > I know what caused it. It's a "test suite" timeout, but the runner > still attempts to execute any remaining tests, even