[jira] [Commented] (LUCENE-8692) IndexWriter.getTragicException() nay not reflect all corrupting exceptions (notably: NoSuchFileException)
[ https://issues.apache.org/jira/browse/LUCENE-8692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16785941#comment-16785941 ] Hoss Man commented on LUCENE-8692: -- {quote}I think we need to be careful here. From my perspective there are 3 types of exceptions here: {quote} Right ok. This is definitely more nuanced then I originally realized, I was lumping #2 and #3 into the same bucket in my head, and hadn't even remembered/considered the "rollback" situation. FWIW: My "mental model" when I first raised this issue was: * We've got this MockDirectoryWrapper.corruptFiles helper method that corrupts files under the covers in ways that are unrecoverable from a client perspective * Presumably then the exceptions a client gets as a result should be "tragic" * When a test uses corruptFiles, here are the code paths that can result in exceptions bubbling up to a client that are not marked tragic But your (IIUC) point is that in the general case, there is no reason some of these code paths (like maybeMerge & startCommit) can/should assume that _any_ exceptions bubbling up are in fact unrecoverable / traggic – in this test it may be, but the code doesn't know that: it's very possible that a rollback will work. {quote}Now we can debate if we want to change this and we can, in-fact I am all for making it even more strict especially since it's inconsistent with what we do if addDocument fails with an aborting exception. {quote} It definitely seems like there should be _something_ we can/should do to better recognize situations like this as "unrecoverable" and be more strict in dealing with low level exceptions during things like commit – but I'm out definitely out of my depth in understanding/suggesting what that might look like. Clearly the current patch is being too aggressive in what it treats as "tragic". > IndexWriter.getTragicException() nay not reflect all corrupting exceptions > (notably: NoSuchFileException) > - > > Key: LUCENE-8692 > URL: https://issues.apache.org/jira/browse/LUCENE-8692 > Project: Lucene - Core > Issue Type: Bug >Reporter: Hoss Man >Priority: Major > Attachments: LUCENE-8692.patch, LUCENE-8692.patch, LUCENE-8692.patch, > LUCENE-8692_test.patch > > > Backstory... > Solr has a "LeaderTragicEventTest" which uses MockDirectoryWrapper's > {{corruptFiles}} to introduce corruption into the "leader" node's index and > then assert that this solr node gives up it's leadership of the shard and > another replica takes over. > This can currently fail sporadically (but usually reproducibly - > seeSOLR-13237) due to the leader not giving up it's leadership even after the > corruption causes an update/commit to fail. Solr's leadership code makes > this decision after encountering an exception from the IndexWriter based on > wether {{IndexWriter.getTragicException()}} is (non-)null. > > While investigating this, I created an isolated Lucene-Core equivilent test > that demonstrates the same basic situation: > * Gradually cause corruption on an index untill (otherwise) valid execution > of IW.add() + IW.commit() calls throw an exception to the IW client. > * assert that if an exception is thrown to the IW client, > {{getTragicException()}} is now non-null. > It's fairly easy to make my new test fail reproducibly -- in every situation > I've seen the underlying exception is a {{NoSuchFileException}} (ie: the > randomly introduced corruption was to delete some file). -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-8692) IndexWriter.getTragicException() nay not reflect all corrupting exceptions (notably: NoSuchFileException)
[ https://issues.apache.org/jira/browse/LUCENE-8692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16785625#comment-16785625 ] Simon Willnauer commented on LUCENE-8692: - {quote} For now I've updated the patch to take the simplest possible approach to checking for MergeAbortedException {quote} +1 {quote} Well, to flip your question around: is there an example of a Throwable you can think of bubbling up out of IndexWriter.startCommit() that should NOT be considered fatal? {quote} I think we need to be careful here. From my perspective there are 3 types of exceptions here: * unrecoverable exceptions aka. VirtualMachineErrors * exceptions that happen during indexing and are not recoverable (these are handled in DocumentsWriter) * exceptions that cause dataloss or inconsistencies (we didn't handle those as fatal yet at least not consistently) but we only catch VirtualMachineError. Those are in particular: * getReader() * deleteAll() * addIndexes() * flushNextBuffer() * prepareCommitInternal() * doFlush() * startCommit() Those methods might cause documents go missing etc. but we treated them not as fatal or tragic events since a user could always call rollback() to go back the the last known safe-point / previous commit. Now we can debate if we want to change this and we can, in-fact I am all for making it even more strict especially since it's inconsistent with what we do if addDocument fails with an aborting exception. If we do that we need to see if rollback still has a purpose and maybe remove it? now speaking of maybeMerge I don't see why we need to close the index writer with a tragic event, there is no dataloss nor an inconsistency? From that logic I don't think we need to handle these exceptions in such a drastic way? {quote} I don't use github for lucene development – I track all contributions as patches in the official issue tracker for the project as recommended by our official guidelines : ) ... but i'll go ahead and create a jira/LUCENE-8692 branch if that will help you review. {quote} Bummer, I am not sure branches help. Working like it's still 1999 is a pain we should fix our guidelines. > IndexWriter.getTragicException() nay not reflect all corrupting exceptions > (notably: NoSuchFileException) > - > > Key: LUCENE-8692 > URL: https://issues.apache.org/jira/browse/LUCENE-8692 > Project: Lucene - Core > Issue Type: Bug >Reporter: Hoss Man >Priority: Major > Attachments: LUCENE-8692.patch, LUCENE-8692.patch, LUCENE-8692.patch, > LUCENE-8692_test.patch > > > Backstory... > Solr has a "LeaderTragicEventTest" which uses MockDirectoryWrapper's > {{corruptFiles}} to introduce corruption into the "leader" node's index and > then assert that this solr node gives up it's leadership of the shard and > another replica takes over. > This can currently fail sporadically (but usually reproducibly - > seeSOLR-13237) due to the leader not giving up it's leadership even after the > corruption causes an update/commit to fail. Solr's leadership code makes > this decision after encountering an exception from the IndexWriter based on > wether {{IndexWriter.getTragicException()}} is (non-)null. > > While investigating this, I created an isolated Lucene-Core equivilent test > that demonstrates the same basic situation: > * Gradually cause corruption on an index untill (otherwise) valid execution > of IW.add() + IW.commit() calls throw an exception to the IW client. > * assert that if an exception is thrown to the IW client, > {{getTragicException()}} is now non-null. > It's fairly easy to make my new test fail reproducibly -- in every situation > I've seen the underlying exception is a {{NoSuchFileException}} (ie: the > randomly introduced corruption was to delete some file). -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-8692) IndexWriter.getTragicException() nay not reflect all corrupting exceptions (notably: NoSuchFileException)
[ https://issues.apache.org/jira/browse/LUCENE-8692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16785006#comment-16785006 ] ASF subversion and git services commented on LUCENE-8692: - Commit 4bd7d6810dbf50ae8f07c1051378f482ff953073 in lucene-solr's branch refs/heads/jira/LUCENE-8692 from Chris M. Hostetter [ https://gitbox.apache.org/repos/asf?p=lucene-solr.git;h=4bd7d68 ] LUCENE-8692: initial attempt at fixing bug > IndexWriter.getTragicException() nay not reflect all corrupting exceptions > (notably: NoSuchFileException) > - > > Key: LUCENE-8692 > URL: https://issues.apache.org/jira/browse/LUCENE-8692 > Project: Lucene - Core > Issue Type: Bug >Reporter: Hoss Man >Priority: Major > Attachments: LUCENE-8692.patch, LUCENE-8692.patch, LUCENE-8692.patch, > LUCENE-8692_test.patch > > > Backstory... > Solr has a "LeaderTragicEventTest" which uses MockDirectoryWrapper's > {{corruptFiles}} to introduce corruption into the "leader" node's index and > then assert that this solr node gives up it's leadership of the shard and > another replica takes over. > This can currently fail sporadically (but usually reproducibly - > seeSOLR-13237) due to the leader not giving up it's leadership even after the > corruption causes an update/commit to fail. Solr's leadership code makes > this decision after encountering an exception from the IndexWriter based on > wether {{IndexWriter.getTragicException()}} is (non-)null. > > While investigating this, I created an isolated Lucene-Core equivilent test > that demonstrates the same basic situation: > * Gradually cause corruption on an index untill (otherwise) valid execution > of IW.add() + IW.commit() calls throw an exception to the IW client. > * assert that if an exception is thrown to the IW client, > {{getTragicException()}} is now non-null. > It's fairly easy to make my new test fail reproducibly -- in every situation > I've seen the underlying exception is a {{NoSuchFileException}} (ie: the > randomly introduced corruption was to delete some file). -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-8692) IndexWriter.getTragicException() nay not reflect all corrupting exceptions (notably: NoSuchFileException)
[ https://issues.apache.org/jira/browse/LUCENE-8692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16785005#comment-16785005 ] ASF subversion and git services commented on LUCENE-8692: - Commit 2b28d9d9d0cdc35dfdc202f23a9ef1d27b2a6083 in lucene-solr's branch refs/heads/jira/LUCENE-8692 from Chris M. Hostetter [ https://gitbox.apache.org/repos/asf?p=lucene-solr.git;h=2b28d9d ] LUCENE-8692: add additional logging to Solr test that first surfaced bug > IndexWriter.getTragicException() nay not reflect all corrupting exceptions > (notably: NoSuchFileException) > - > > Key: LUCENE-8692 > URL: https://issues.apache.org/jira/browse/LUCENE-8692 > Project: Lucene - Core > Issue Type: Bug >Reporter: Hoss Man >Priority: Major > Attachments: LUCENE-8692.patch, LUCENE-8692.patch, LUCENE-8692.patch, > LUCENE-8692_test.patch > > > Backstory... > Solr has a "LeaderTragicEventTest" which uses MockDirectoryWrapper's > {{corruptFiles}} to introduce corruption into the "leader" node's index and > then assert that this solr node gives up it's leadership of the shard and > another replica takes over. > This can currently fail sporadically (but usually reproducibly - > seeSOLR-13237) due to the leader not giving up it's leadership even after the > corruption causes an update/commit to fail. Solr's leadership code makes > this decision after encountering an exception from the IndexWriter based on > wether {{IndexWriter.getTragicException()}} is (non-)null. > > While investigating this, I created an isolated Lucene-Core equivilent test > that demonstrates the same basic situation: > * Gradually cause corruption on an index untill (otherwise) valid execution > of IW.add() + IW.commit() calls throw an exception to the IW client. > * assert that if an exception is thrown to the IW client, > {{getTragicException()}} is now non-null. > It's fairly easy to make my new test fail reproducibly -- in every situation > I've seen the underlying exception is a {{NoSuchFileException}} (ie: the > randomly introduced corruption was to delete some file). -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-8692) IndexWriter.getTragicException() nay not reflect all corrupting exceptions (notably: NoSuchFileException)
[ https://issues.apache.org/jira/browse/LUCENE-8692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16785004#comment-16785004 ] ASF subversion and git services commented on LUCENE-8692: - Commit af9da98e87ee2a44532cd9baf2a1bc6e64865b0d in lucene-solr's branch refs/heads/jira/LUCENE-8692 from Chris M. Hostetter [ https://gitbox.apache.org/repos/asf?p=lucene-solr.git;h=af9da98 ] LUCENE-8692: updates to TestStressIndexing2 demonstrating bug > IndexWriter.getTragicException() nay not reflect all corrupting exceptions > (notably: NoSuchFileException) > - > > Key: LUCENE-8692 > URL: https://issues.apache.org/jira/browse/LUCENE-8692 > Project: Lucene - Core > Issue Type: Bug >Reporter: Hoss Man >Priority: Major > Attachments: LUCENE-8692.patch, LUCENE-8692.patch, LUCENE-8692.patch, > LUCENE-8692_test.patch > > > Backstory... > Solr has a "LeaderTragicEventTest" which uses MockDirectoryWrapper's > {{corruptFiles}} to introduce corruption into the "leader" node's index and > then assert that this solr node gives up it's leadership of the shard and > another replica takes over. > This can currently fail sporadically (but usually reproducibly - > seeSOLR-13237) due to the leader not giving up it's leadership even after the > corruption causes an update/commit to fail. Solr's leadership code makes > this decision after encountering an exception from the IndexWriter based on > wether {{IndexWriter.getTragicException()}} is (non-)null. > > While investigating this, I created an isolated Lucene-Core equivilent test > that demonstrates the same basic situation: > * Gradually cause corruption on an index untill (otherwise) valid execution > of IW.add() + IW.commit() calls throw an exception to the IW client. > * assert that if an exception is thrown to the IW client, > {{getTragicException()}} is now non-null. > It's fairly easy to make my new test fail reproducibly -- in every situation > I've seen the underlying exception is a {{NoSuchFileException}} (ie: the > randomly introduced corruption was to delete some file). -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-8692) IndexWriter.getTragicException() nay not reflect all corrupting exceptions (notably: NoSuchFileException)
[ https://issues.apache.org/jira/browse/LUCENE-8692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16785007#comment-16785007 ] ASF subversion and git services commented on LUCENE-8692: - Commit e93313d42c4e5642e58707b74ed119776086b590 in lucene-solr's branch refs/heads/jira/LUCENE-8692 from Chris M. Hostetter [ https://gitbox.apache.org/repos/asf?p=lucene-solr.git;h=e93313d ] LUCENE-8692: fix file leak in MockRandomPostingsFormat that surfaces when corrupting files > IndexWriter.getTragicException() nay not reflect all corrupting exceptions > (notably: NoSuchFileException) > - > > Key: LUCENE-8692 > URL: https://issues.apache.org/jira/browse/LUCENE-8692 > Project: Lucene - Core > Issue Type: Bug >Reporter: Hoss Man >Priority: Major > Attachments: LUCENE-8692.patch, LUCENE-8692.patch, LUCENE-8692.patch, > LUCENE-8692_test.patch > > > Backstory... > Solr has a "LeaderTragicEventTest" which uses MockDirectoryWrapper's > {{corruptFiles}} to introduce corruption into the "leader" node's index and > then assert that this solr node gives up it's leadership of the shard and > another replica takes over. > This can currently fail sporadically (but usually reproducibly - > seeSOLR-13237) due to the leader not giving up it's leadership even after the > corruption causes an update/commit to fail. Solr's leadership code makes > this decision after encountering an exception from the IndexWriter based on > wether {{IndexWriter.getTragicException()}} is (non-)null. > > While investigating this, I created an isolated Lucene-Core equivilent test > that demonstrates the same basic situation: > * Gradually cause corruption on an index untill (otherwise) valid execution > of IW.add() + IW.commit() calls throw an exception to the IW client. > * assert that if an exception is thrown to the IW client, > {{getTragicException()}} is now non-null. > It's fairly easy to make my new test fail reproducibly -- in every situation > I've seen the underlying exception is a {{NoSuchFileException}} (ie: the > randomly introduced corruption was to delete some file). -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-8692) IndexWriter.getTragicException() nay not reflect all corrupting exceptions (notably: NoSuchFileException)
[ https://issues.apache.org/jira/browse/LUCENE-8692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16784997#comment-16784997 ] Hoss Man commented on LUCENE-8692: -- bq. (see also my nocommit comments about the existing tragicEvent() call in prepareCommitInternal() ... but that hadn't triggered any failures in the test so I hadn't touched it) I spoke too soon -- beasting just turned up this interesting little situation... {noformat} hossman@tray:~/lucene/dev/lucene/core [master] $ ant beast -Dbeast.iters=100 -Dtests.iters=100 -Dtestcase=TestStressIndexing2 -Dtests.method=testRandomCorruptionIsTragic\* ... [beaster] Beast round 34 results: /home/hossman/lucene/dev/lucene/build/core/test/34 [beaster] The following error occurred while executing this line: [beaster] /home/hossman/lucene/dev/lucene/common-build.xml:1572: The following error occurred while executing this line: [beaster] /home/hossman/lucene/dev/lucene/common-build.xml:1099: There were test failures: 1 suite, 100 tests, 1 failure [seed: CABE666E4674CFB2] [beaster] Executing 1 suite with 1 JVM. [beaster] [beaster] Started J0 PID(10111@localhost). [beaster] 2> NOTE: reproduce with: ant test -Dtestcase=TestStressIndexing2 -Dtests.method=testRandomCorruptionIsTragic -Dtests.seed=CABE666E4674CFB2 -Dtests.slow=true -Dtests.badapples=true -Dtests.locale=cs -Dtests.timezone=America/Nipigon -Dtests.asserts=true -Dtests.file.encoding=UTF-8 [beaster] [15:50:16.736] FAILURE 0.02s | TestStressIndexing2.testRandomCorruptionIsTragic {seed=[CABE666E4674CFB2:682DC0F2BA2A235F]} <<< [beaster]> Throwable #1: java.lang.AssertionError: index update encountered throwable, but no tragic event recorded: java.lang.AssertionError [beaster]>at __randomizedtesting.SeedInfo.seed([CABE666E4674CFB2:682DC0F2BA2A235F]:0) [beaster]>at org.junit.Assert.fail(Assert.java:88) [beaster]>at org.junit.Assert.assertTrue(Assert.java:41) [beaster]>at org.junit.Assert.assertNotNull(Assert.java:712) [beaster]>at org.apache.lucene.index.TestStressIndexing2$CorruptibleIndexingThread.run(TestStressIndexing2.java:1019) [beaster]>at org.apache.lucene.index.TestStressIndexing2.testRandomCorruptionIsTragic(TestStressIndexing2.java:144) [beaster]>at sun.reflect.GeneratedMethodAccessor3.invoke(Unknown Source) [beaster]>at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43) [beaster]>at java.lang.reflect.Method.invoke(Method.java:498) [beaster]>at com.carrotsearch.randomizedtesting.RandomizedRunner.invoke(RandomizedRunner.java:1750) [beaster]>at com.carrotsearch.randomizedtesting.RandomizedRunner$8.evaluate(RandomizedRunner.java:938) [beaster]>at com.carrotsearch.randomizedtesting.RandomizedRunner$9.evaluate(RandomizedRunner.java:974) [beaster]>at com.carrotsearch.randomizedtesting.RandomizedRunner$10.evaluate(RandomizedRunner.java:988) [beaster]>at org.apache.lucene.util.TestRuleSetupTeardownChained$1.evaluate(TestRuleSetupTeardownChained.java:49) [beaster]>at org.apache.lucene.util.AbstractBeforeAfterRule$1.evaluate(AbstractBeforeAfterRule.java:45) [beaster]>at org.apache.lucene.util.TestRuleThreadAndTestName$1.evaluate(TestRuleThreadAndTestName.java:48) [beaster]>at org.apache.lucene.util.TestRuleIgnoreAfterMaxFailures$1.evaluate(TestRuleIgnoreAfterMaxFailures.java:64) [beaster]>at org.apache.lucene.util.TestRuleMarkFailure$1.evaluate(TestRuleMarkFailure.java:47) [beaster]>at com.carrotsearch.randomizedtesting.rules.StatementAdapter.evaluate(StatementAdapter.java:36) [beaster]>at com.carrotsearch.randomizedtesting.ThreadLeakControl$StatementRunner.run(ThreadLeakControl.java:368) [beaster]>at com.carrotsearch.randomizedtesting.ThreadLeakControl.forkTimeoutingTask(ThreadLeakControl.java:817) [beaster]>at com.carrotsearch.randomizedtesting.ThreadLeakControl$3.evaluate(ThreadLeakControl.java:468) [beaster]>at com.carrotsearch.randomizedtesting.RandomizedRunner.runSingleTest(RandomizedRunner.java:947) [beaster]>at com.carrotsearch.randomizedtesting.RandomizedRunner$5.evaluate(RandomizedRunner.java:832) [beaster]>at com.carrotsearch.randomizedtesting.RandomizedRunner$6.evaluate(RandomizedRunner.java:883) [beaster]>at com.carrotsearch.randomizedtesting.RandomizedRunner$7.evaluate(RandomizedRunner.java:894) [beaster]>at org.apache.lucene.util.AbstractBeforeAfterRule$1.evaluate(AbstractBeforeAfterRule.java:45) [beaster]>at com.carrotsearch.randomizedtesting.rules.StatementAdapter.evaluate(StatementAdapter.java:36) [beaster]>at
[jira] [Commented] (LUCENE-8692) IndexWriter.getTragicException() nay not reflect all corrupting exceptions (notably: NoSuchFileException)
[ https://issues.apache.org/jira/browse/LUCENE-8692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16784438#comment-16784438 ] Simon Willnauer commented on LUCENE-8692: - {noformat} I think there is an issue with the patch with MergeAbortedExeption indeed given that registerMerge might throw such an exception. Maybe we should move this try block to registerMerge instead where we know which OneMerge is being registered (and is also where the exception is thrown when estimating the size of the merge). {noformat} +1 {code:java} -} catch (VirtualMachineError tragedy) { +} catch (Throwable tragedy) { tragicEvent(tragedy, "startCommit"); {code} I am not sure why we need to treat every exception as fatal in this case? I also wonder if we could move this to a PR on github, iterations would be simpler and comments too. I can't tell which patch is relevant which one isn't. > IndexWriter.getTragicException() nay not reflect all corrupting exceptions > (notably: NoSuchFileException) > - > > Key: LUCENE-8692 > URL: https://issues.apache.org/jira/browse/LUCENE-8692 > Project: Lucene - Core > Issue Type: Bug >Reporter: Hoss Man >Priority: Major > Attachments: LUCENE-8692.patch, LUCENE-8692.patch, > LUCENE-8692_test.patch > > > Backstory... > Solr has a "LeaderTragicEventTest" which uses MockDirectoryWrapper's > {{corruptFiles}} to introduce corruption into the "leader" node's index and > then assert that this solr node gives up it's leadership of the shard and > another replica takes over. > This can currently fail sporadically (but usually reproducibly - > seeSOLR-13237) due to the leader not giving up it's leadership even after the > corruption causes an update/commit to fail. Solr's leadership code makes > this decision after encountering an exception from the IndexWriter based on > wether {{IndexWriter.getTragicException()}} is (non-)null. > > While investigating this, I created an isolated Lucene-Core equivilent test > that demonstrates the same basic situation: > * Gradually cause corruption on an index untill (otherwise) valid execution > of IW.add() + IW.commit() calls throw an exception to the IW client. > * assert that if an exception is thrown to the IW client, > {{getTragicException()}} is now non-null. > It's fairly easy to make my new test fail reproducibly -- in every situation > I've seen the underlying exception is a {{NoSuchFileException}} (ie: the > randomly introduced corruption was to delete some file). -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-8692) IndexWriter.getTragicException() nay not reflect all corrupting exceptions (notably: NoSuchFileException)
[ https://issues.apache.org/jira/browse/LUCENE-8692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16783151#comment-16783151 ] Adrien Grand commented on LUCENE-8692: -- bq. the new test seems to have also uncovered some issues with IndexWriter leaking files in some tragic event situations This one is a bug in {{MockRandomPostingsFormat#fieldsProducer}}. It should use try-with-resources when opening {{in}} to make sure it eventually gets closed. I think there is an issue with the patch with MergeAbortedExeption indeed given that {{registerMerge}} might throw such an exception. Maybe we should move this try block to {{registerMerge}} instead where we know which OneMerge is being registered (and is also where the exception is thrown when estimating the size of the merge). [~mikemccand] [~rcmuir] [~simonw] You might want to have a look at this patch since you are more familiar with tragedy handling than I am. > IndexWriter.getTragicException() nay not reflect all corrupting exceptions > (notably: NoSuchFileException) > - > > Key: LUCENE-8692 > URL: https://issues.apache.org/jira/browse/LUCENE-8692 > Project: Lucene - Core > Issue Type: Bug >Reporter: Hoss Man >Priority: Major > Attachments: LUCENE-8692.patch, LUCENE-8692.patch, > LUCENE-8692_test.patch > > > Backstory... > Solr has a "LeaderTragicEventTest" which uses MockDirectoryWrapper's > {{corruptFiles}} to introduce corruption into the "leader" node's index and > then assert that this solr node gives up it's leadership of the shard and > another replica takes over. > This can currently fail sporadically (but usually reproducibly - > seeSOLR-13237) due to the leader not giving up it's leadership even after the > corruption causes an update/commit to fail. Solr's leadership code makes > this decision after encountering an exception from the IndexWriter based on > wether {{IndexWriter.getTragicException()}} is (non-)null. > > While investigating this, I created an isolated Lucene-Core equivilent test > that demonstrates the same basic situation: > * Gradually cause corruption on an index untill (otherwise) valid execution > of IW.add() + IW.commit() calls throw an exception to the IW client. > * assert that if an exception is thrown to the IW client, > {{getTragicException()}} is now non-null. > It's fairly easy to make my new test fail reproducibly -- in every situation > I've seen the underlying exception is a {{NoSuchFileException}} (ie: the > randomly introduced corruption was to delete some file). -- This message was sent by Atlassian JIRA (v7.6.3#76005) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (LUCENE-8692) IndexWriter.getTragicException() nay not reflect all corrupting exceptions (notably: NoSuchFileException)
[ https://issues.apache.org/jira/browse/LUCENE-8692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16766039#comment-16766039 ] Adrien Grand commented on LUCENE-8692: -- I had a quick look and every time the test failed for me it was in the maybeMerge call that is triggered by the commit() call, eg. {noformat} java.nio.file.NoSuchFileException: _0.fdx at org.apache.lucene.store.ByteBuffersDirectory.fileLength(ByteBuffersDirectory.java:155) at org.apache.lucene.store.MockDirectoryWrapper.fileLength(MockDirectoryWrapper.java:1009) at org.apache.lucene.index.SegmentCommitInfo.sizeInBytes(SegmentCommitInfo.java:217) at org.apache.lucene.index.IndexWriter.registerMerge(IndexWriter.java:4176) at org.apache.lucene.index.IndexWriter.updatePendingMerges(IndexWriter.java:2197) at org.apache.lucene.index.IndexWriter.maybeMerge(IndexWriter.java:2154) at org.apache.lucene.index.IndexWriter.commitInternal(IndexWriter.java:3455) at org.apache.lucene.index.IndexWriter.commit(IndexWriter.java:3407) at org.apache.lucene.index.TestIndexWriterExceptions.testRandomCorruptionIsTragic(TestIndexWriterExceptions.java:2136) at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62) at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43) at java.lang.reflect.Method.invoke(Method.java:498) at com.carrotsearch.randomizedtesting.RandomizedRunner.invoke(RandomizedRunner.java:1750) at com.carrotsearch.randomizedtesting.RandomizedRunner$8.evaluate(RandomizedRunner.java:938) at com.carrotsearch.randomizedtesting.RandomizedRunner$9.evaluate(RandomizedRunner.java:974) at com.carrotsearch.randomizedtesting.RandomizedRunner$10.evaluate(RandomizedRunner.java:988) at org.apache.lucene.util.TestRuleSetupTeardownChained$1.evaluate(TestRuleSetupTeardownChained.java:49) at org.apache.lucene.util.AbstractBeforeAfterRule$1.evaluate(AbstractBeforeAfterRule.java:45) at org.apache.lucene.util.TestRuleThreadAndTestName$1.evaluate(TestRuleThreadAndTestName.java:48) at org.apache.lucene.util.TestRuleIgnoreAfterMaxFailures$1.evaluate(TestRuleIgnoreAfterMaxFailures.java:64) at org.apache.lucene.util.TestRuleMarkFailure$1.evaluate(TestRuleMarkFailure.java:47) at com.carrotsearch.randomizedtesting.rules.StatementAdapter.evaluate(StatementAdapter.java:36) at com.carrotsearch.randomizedtesting.ThreadLeakControl$StatementRunner.run(ThreadLeakControl.java:368) at com.carrotsearch.randomizedtesting.ThreadLeakControl.forkTimeoutingTask(ThreadLeakControl.java:817) at com.carrotsearch.randomizedtesting.ThreadLeakControl$3.evaluate(ThreadLeakControl.java:468) at com.carrotsearch.randomizedtesting.RandomizedRunner.runSingleTest(RandomizedRunner.java:947) at com.carrotsearch.randomizedtesting.RandomizedRunner$5.evaluate(RandomizedRunner.java:832) at com.carrotsearch.randomizedtesting.RandomizedRunner$6.evaluate(RandomizedRunner.java:883) at com.carrotsearch.randomizedtesting.RandomizedRunner$7.evaluate(RandomizedRunner.java:894) at org.apache.lucene.util.AbstractBeforeAfterRule$1.evaluate(AbstractBeforeAfterRule.java:45) at com.carrotsearch.randomizedtesting.rules.StatementAdapter.evaluate(StatementAdapter.java:36) at org.apache.lucene.util.TestRuleStoreClassName$1.evaluate(TestRuleStoreClassName.java:41) at com.carrotsearch.randomizedtesting.rules.NoShadowingOrOverridesOnMethodsRule$1.evaluate(NoShadowingOrOverridesOnMethodsRule.java:40) at com.carrotsearch.randomizedtesting.rules.NoShadowingOrOverridesOnMethodsRule$1.evaluate(NoShadowingOrOverridesOnMethodsRule.java:40) at com.carrotsearch.randomizedtesting.rules.StatementAdapter.evaluate(StatementAdapter.java:36) at com.carrotsearch.randomizedtesting.rules.StatementAdapter.evaluate(StatementAdapter.java:36) at com.carrotsearch.randomizedtesting.rules.StatementAdapter.evaluate(StatementAdapter.java:36) at org.apache.lucene.util.TestRuleAssertionsRequired$1.evaluate(TestRuleAssertionsRequired.java:53) at org.apache.lucene.util.TestRuleMarkFailure$1.evaluate(TestRuleMarkFailure.java:47) at org.apache.lucene.util.TestRuleIgnoreAfterMaxFailures$1.evaluate(TestRuleIgnoreAfterMaxFailures.java:64) at org.apache.lucene.util.TestRuleIgnoreTestSuites$1.evaluate(TestRuleIgnoreTestSuites.java:54) at com.carrotsearch.randomizedtesting.rules.StatementAdapter.evaluate(StatementAdapter.java:36) at com.carrotsearch.randomizedtesting.ThreadLeakControl$StatementRunner.run(ThreadLeakControl.java:368) at java.lang.Thread.run(Thread.java:748) {noformat}