Community
Participate
Working Groups
Many (indeed most) of our offline test are currently failing. Some issues I've already identified: 1. Expected revision counts are off due to use of test-specific resource folder (#getResourcePath). 2. TimestampAuthority.endCommit fails to notify threads suspended in waitForCommit. Investigating further.
3. Something is invoking raw replication for MEMStore, which doesn't support raw mode. This causes an endless string of OpNotSupportedEx's, which never bubble up to the client, it seems.
@3.. my mistake. I enabled the raw tests for the MEMStore explicitly.
4. An apparent bug in AbstractSyncingTest.checkEvent. The PollingTimeouter checks the length of the events array. But if the condition fails the first time, how could it ever pass later? The (growing) list of events inside the TestListener is converted to an array only once; of course, the length of this array will not change later.
5. Bug 349804
6. In the logic that creates a ChangeSetData instance after raw replication, it appears to have been overlooked that the constructed delta could (in fact: is likely to) represent a compound change; that is, a change from rev m to rev n, where n > m+1. The delta ends up being applied to the clone's revision m, raising its revision number with +1, while in fact the state conveyed belongs to a later revision. When the client makes later modifications to that revision and commits them, a "attempt to modify historical revision" gets thrown. This causes the failure of testSynchronizationMasterCloneWithReplication.
@6. Ultimately the client sets the version of the new revision (which carries the state conveyed by the delta) by calling adjustForCommit (AbstractCDORevision). This is where the +1 assumption is hardcoded.
Created attachment 198812 [details] Patch v1
Patch v1 addresses issues 1, 2, 3, 4 and 5.
Created attachment 198872 [details] Patch v2
Patch v2 addresses 1 through 6, as well as: 7. CDOStoreImpl.set calls getRevision where it should use getRevisionForReading. The former does not load the revision if necessary, so that an NPE arises if a setter is called on an object in PROXY state.
There are some problems with the examples as well. 8. After starting up the OfflineExampleMaster, OfflineExampleClone, and OfflineExampleClient (in that order), it turns out that the clone has returned a rootresourceID of NULL to the client. When the client actually tries to perform some ops, an NPE follows.
Changing to 4.1 to ensure that the fix will "last". Please clone this bugzilla to 4.0 if you want a maintenance fix, too.
Note: your changes in TimeStampAuthority will partially conflict with a recent fix from Egidijus. IIRC Egidijus had the same fix for *parts* of your changes.
Can it be that you forgot to revert your changes in AllTestsMEMOffline? See your comment #2 ...
Go ahead and commit to *trunk* . Is it true that we're now down to 16/1 failures out of 34 tests? What about problem 8 in comment #11 ?
Committed revision 8579.
Cloned for 4.0 as Bug 351046
(In reply to comment #15) > Is it true that we're now down to 16/1 failures out of 34 tests? No. I get 0 errors, 1 failure out of 34 tests. And that single failure only happens when that test is run as part of the suite. When I re-run it separately, it always passes. :-S > What about problem 8 in comment #11 ? I think I caused the problem myself by debugging multiple threads simultaneously and suspending the thread that was supposed to set the rootResourceID -- not 100% sure. :S
Strange, every test case n (n is an odd number) passes but blocks for some while in the teardown process. Every next test case n+1 fails in the setup process: org.eclipse.emf.cdo.tests.config.impl.ConfigTestException: Error in OfflineTest.testClientCommits [Combined, DBStore: H2 (offline), JVM, Native] at org.eclipse.emf.cdo.tests.config.impl.ConfigTest.runBare(ConfigTest.java:516) at junit.framework.TestResult$1.protect(TestResult.java:110) at junit.framework.TestResult.runProtected(TestResult.java:128) at junit.framework.TestResult.run(TestResult.java:113) at junit.framework.TestCase.run(TestCase.java:124) at org.eclipse.net4j.util.tests.AbstractOMTest.run(AbstractOMTest.java:260) at junit.framework.TestSuite.runTest(TestSuite.java:243) at org.eclipse.emf.cdo.tests.config.impl.ConfigTestSuite$TestWrapper.runTest(ConfigTestSuite.java:126) at junit.framework.TestSuite.run(TestSuite.java:238) at junit.framework.TestSuite.runTest(TestSuite.java:243) at junit.framework.TestSuite.run(TestSuite.java:238) at junit.framework.TestSuite.runTest(TestSuite.java:243) at junit.framework.TestSuite.run(TestSuite.java:238) at org.eclipse.jdt.internal.junit.runner.junit3.JUnit3TestReference.run(JUnit3TestReference.java:130) at org.eclipse.jdt.internal.junit.runner.TestExecution.run(TestExecution.java:38) at org.eclipse.jdt.internal.junit.runner.RemoteTestRunner.runTests(RemoteTestRunner.java:467) at org.eclipse.jdt.internal.junit.runner.RemoteTestRunner.runTests(RemoteTestRunner.java:683) at org.eclipse.jdt.internal.junit.runner.RemoteTestRunner.run(RemoteTestRunner.java:390) at org.eclipse.jdt.internal.junit.runner.RemoteTestRunner.main(RemoteTestRunner.java:197) Caused by: java.lang.IllegalStateException: Not active: ManagedContainer at org.eclipse.net4j.util.lifecycle.LifecycleUtil.checkActive(LifecycleUtil.java:72) at org.eclipse.net4j.util.lifecycle.Lifecycle.checkActive(Lifecycle.java:190) at org.eclipse.net4j.util.container.ManagedContainer.getElement(ManagedContainer.java:273) at org.eclipse.net4j.util.container.ManagedContainer.getElement(ManagedContainer.java:265) at org.eclipse.net4j.jvm.JVMUtil.getAcceptor(JVMUtil.java:36) at org.eclipse.emf.cdo.tests.config.impl.SessionConfig$Net4j$JVM.getAcceptor(SessionConfig.java:492) at org.eclipse.emf.cdo.tests.config.impl.SessionConfig$Net4j.startTransport(SessionConfig.java:253) at org.eclipse.emf.cdo.tests.config.impl.ConfigTest.startTransport(ConfigTest.java:279) at org.eclipse.emf.cdo.tests.AbstractCDOTest.doSetUp(AbstractCDOTest.java:57) at org.eclipse.net4j.util.tests.AbstractOMTest.setUp(AbstractOMTest.java:159) at org.eclipse.emf.cdo.tests.config.impl.ConfigTest.setUp(ConfigTest.java:694) at org.eclipse.net4j.util.tests.AbstractOMTest.runBare(AbstractOMTest.java:213) at org.eclipse.emf.cdo.tests.config.impl.ConfigTest.runBare(ConfigTest.java:507) ... 18 more
This is the stack trace of the main thread when blocking some time: Thread [main] (Suspended) owns: Scenario (id=77) waiting for: Worker$WorkerThread (id=38) Object.wait(long) line: not available [native method] Worker$WorkerThread(Thread).join(long) line: 1194 RepositorySynchronizer(Worker).doDeactivate() line: 120 RepositorySynchronizer(QueueWorker<E>).doDeactivate() line: 110 RepositorySynchronizer.doDeactivate() line: 206 RepositorySynchronizer(Lifecycle).deactivate() line: 125 RepositoryConfig$OfflineConfig$1(SynchronizableRepository).stopSynchronization() line: 402 RepositoryConfig$OfflineConfig$1(SynchronizableRepository).doDeactivate() line: 376 RepositoryConfig$OfflineConfig$1(Lifecycle).deactivate() line: 125 LifecycleUtil.deactivate(Object, boolean) line: 206 LifecycleUtil.deactivate(Object) line: 196 AllTestsDBH2Offline$H2Offline$ReusableFolder(RepositoryConfig).deactivateRepositories() line: 318 AllTestsDBH2Offline$H2Offline$ReusableFolder(RepositoryConfig$OfflineConfig).deactivateRepositories() line: 558 AllTestsDBH2Offline$H2Offline$ReusableFolder(RepositoryConfig).tearDown() line: 284 Scenario.tearDown() line: 230 OfflineTest(ConfigTest).doTearDown() line: 709 OfflineTest(AbstractCDOTest).doTearDown() line: 66 OfflineTest(AbstractOMTest).tearDown() line: 180 OfflineTest(AbstractOMTest).runBare() line: 224 OfflineTest(ConfigTest).runBare() line: 507 TestResult$1.protect() line: 110 TestResult.runProtected(Test, Protectable) line: 128 TestResult.run(TestCase) line: 113 OfflineTest(TestCase).run(TestResult) line: 124 OfflineTest(AbstractOMTest).run(TestResult) line: 260 JUnit3TestReference.run(TestExecution) line: 130 TestExecution.run(ITestReference[]) line: 38 RemoteTestRunner.runTests(String[], String, TestExecution) line: 467 RemoteTestRunner.runTests(TestExecution) line: 683 RemoteTestRunner.run() line: 390 RemoteTestRunner.main(String[]) line: 197
Committed revision 8605: - trunk/plugins/org.eclipse.emf.cdo.tests - trunk/plugins/org.eclipse.net4j.util
Okay, I found it. It was a temporary deadlock situation between the main thread and the RepositorSynchronizer thread. After removing the picky synchronized modifiers from all Scenario.java methods I get the same results as you: only 1 failure, and that only if executed in the suite.
I think we can resolve this for now. Thanks!
Closing.
