Am 29.10.14 um 12:25 schrieb Stefan Seifert: > i tried to run the integration tests for the event part on my (windows) > machines, sometimes the pass, sometimes they fail, not always with the same > result. but never because of a port mismatch. > > today I started a new try: > > * windows 7 + JDK 1.7: tests passed (but a showing frightening lot of > exceptions like " javax.jcr.InvalidItemStateException: Item does not exist > anymore: 897fc158-e48e-480c-b5af-cc8d761d959a") > Yes, that's expected - unfortunately. The code is iterating over a tree while some nodes are removed concurrently. WHich results in these noisy messages - but the do no harm. > > * windows 7 + JKD 1.8: > > --- > Failed tests: > ChaosTest.testDoChaos:397 Checking topic sling/chaos/ordered/0 > expected:<3387> but was:<3386>
This is actually a problem with the slowness of the observation handler.Before the observation handler can deliver the event about a resource, the resource is already processed by the job handling. I'll remove this check from the code as it's not testing anything new. > > Tests run: 20, Failures: 1, Errors: 0, Skipped: 0 > --- > > * windows 7 + JKD 1.8 (repeated run without other changes - seems to hang up > somewhere and than run into timeouts): > > --- > Failed tests: > ClassloadingTest.testFailedClassloading:204 RetryLoop failed, condition is > false after 5 seconds: Waiting for job failure to be recorded. Conditions > faildJobsCount=0, finishedEvents=0, findJobs= 1, queuedJobs=0, active > Jobs=0 > ClassloadingTest.testSimpleClassloading:131 RetryLoop failed, condition is > false after 5 seconds: Waiting for finishedEvents to have at least one element > DeprecatedJobsTest.testSimpleJobExecutionUsingBridge:120 No event received > in the given time. > DeprecatedJobsTest.testSimpleJobWithIdExecution:147 No event received in > the given time. > JobHandlingTest.testSimpleJobExecutionUsingJobExecutor:139 No event > received in the given time. > JobHandlingTest.testStartJobAndReschedule:304 No event received in the > given time. > JobHandlingTest.testSimpleJobExecutionUsingJobConsumer:111 No event > received in the given time. > OrderedQueueTest.testOrderedQueue:147 No event received in the given time. > RoundRobinQueueTest.testRoundRobinQueue:132 No event received in the given > time. > > Tests in error: > ChaosTest.testDoChaos:393->AbstractJobHandlingTest.sleep:175 » test timed > out... > DeprecatedJobsTest.testForceCancelJob:175 » test timed out after 300000 > milli... > HistoryTest.testHistory:117 » test timed out after 300000 milliseconds > JobHandlingTest.testNoJobProcessor:472 » test timed out after 300000 > millisec... > JobHandlingTest.testNotifications:422 Timeout during notification test. > JobHandlingTest.testManyJobs:180 » test timed out after 300000 milliseconds > JobHandlingTest.testCancelJob:214 » test timed out after 300000 > milliseconds > JobHandlingTest.testGetJob:262 » test timed out after 300000 milliseconds > TimedJobsTest.testTimedJob:84->AbstractJobHandlingTest.sleep:175 » test > timed... > > Tests run: 20, Failures: 9, Errors: 9, Skipped: 0 > --- > > with all this pax-exam stuff i do not know in detail it's difficult to tell > where it makes sense debugging or further analyzing. > last week i had problems running the tests with JDK 1.7 as well, so i assume > it's not a JDK 1.8 related issue. I think the last one is rather a general problem with pax in that situation. Regards Carsten > > stefan > > > >> -----Original Message----- >> From: Carsten Ziegeler [mailto:[email protected]] >> Sent: Tuesday, October 28, 2014 8:58 PM >> To: [email protected] >> Subject: Re: Jenkins build is still unstable: sling-trunk-1.8 » Apache Sling >> Event Support #381 >> >> Am 28.10.14 um 17:22 schrieb Apache Jenkins Server: >>> See <https://builds.apache.org/job/sling-trunk- >> 1.8/org.apache.sling$org.apache.sling.event/381/> >>> >>> >> This is a sporadic problem with Java 8 - it seems when a new Pax Exam >> test starts, the old one isn't finished yet as the port is still in use >> and the repository is locked. >> >> Does anyone have a good idea how to workaround this problem or let pax >> wait until really all threads are finished? >> >> Or even better, can anyone reproduce this locally so we could get a >> thread dump? >> >> Carsten >> >> -- >> Carsten Ziegeler >> Adobe Research Switzerland >> [email protected] -- Carsten Ziegeler Adobe Research Switzerland [email protected]
