[jira] [Commented] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16802113#comment-16802113 ] Kevin Risden commented on SOLR-5007: Planning to follow up on the HDFS mailing list or something. I reached out to some colleagues about `FileSystem` vs `FileContext` but haven't heard anything. > TestRecoveryHdfs seems to be leaking a thread occasionally that ends up > failing a completely different test. > > > Key: SOLR-5007 > URL: https://issues.apache.org/jira/browse/SOLR-5007 > Project: Solr > Issue Type: Test > Components: Hadoop Integration, hdfs, Tests >Reporter: Mark Miller >Assignee: Kevin Risden >Priority: Major > Attachments: SOLR-5007.patch > > -- 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] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16798304#comment-16798304 ] Kevin Risden commented on SOLR-5007: Putting some notes here as I looked into this a bit more. Hadoop `FileSystem` vs `FileContext`. `FileContext` seems to be the preferred HDFS interaction tool. but `FileContext` doesn’t have a `close()` method so can’t clean up any usages. In the Solr HDFS stuff, there is a mix of `FileSystem` and `FileContext`. and it is weird from a resource usage as to what should be used. I know all the `FileContext` stuff leaks threads in Solr testing, but moving to `FileSystem` isn’t an option since not all the required methods are on `FileSystem`. I plan to try to follow up with the Hadoop ecosystem about `FileContext` and how to properly clean up. Since right now there is no easy way :( > TestRecoveryHdfs seems to be leaking a thread occasionally that ends up > failing a completely different test. > > > Key: SOLR-5007 > URL: https://issues.apache.org/jira/browse/SOLR-5007 > Project: Solr > Issue Type: Test > Components: Hadoop Integration, hdfs, Tests >Reporter: Mark Miller >Assignee: Kevin Risden >Priority: Major > Attachments: SOLR-5007.patch > > -- 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] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16772408#comment-16772408 ] Kevin Risden commented on SOLR-5007: Worked on this a bunch and then got side tracked. I found a few cases where we can clean up our handling of HDFS clients. Looks like Hadoop code still leaks threads even on proper shutdown. > TestRecoveryHdfs seems to be leaking a thread occasionally that ends up > failing a completely different test. > > > Key: SOLR-5007 > URL: https://issues.apache.org/jira/browse/SOLR-5007 > Project: Solr > Issue Type: Test > Components: Hadoop Integration, hdfs, Tests >Reporter: Mark Miller >Assignee: Kevin Risden >Priority: Major > Attachments: SOLR-5007.patch > > -- 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] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16761071#comment-16761071 ] Kevin Risden commented on SOLR-5007: Probably going to have to address this piece meal doing one test at a time to make sure things are cleaned up/closed properly. BadHdfsThreadsFilter is like a hammer and used in all the HDFS tests whether all the threads leak there or not. > TestRecoveryHdfs seems to be leaking a thread occasionally that ends up > failing a completely different test. > > > Key: SOLR-5007 > URL: https://issues.apache.org/jira/browse/SOLR-5007 > Project: Solr > Issue Type: Test > Components: Hadoop Integration, hdfs, Tests >Reporter: Mark Miller >Assignee: Kevin Risden >Priority: Major > Attachments: SOLR-5007.patch > > -- 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] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16760339#comment-16760339 ] Kevin Risden commented on SOLR-5007: Well this uncovered a bunch of other code/tests that aren't doing the right thing. Backup/restore doesn't seem to properly close things. HdfsLockFactory doesn't close the Filesystem it creates. Planning to look at this more but there are some definite resource management issues here that have been covered up for a long time. > TestRecoveryHdfs seems to be leaking a thread occasionally that ends up > failing a completely different test. > > > Key: SOLR-5007 > URL: https://issues.apache.org/jira/browse/SOLR-5007 > Project: Solr > Issue Type: Test > Components: Hadoop Integration, hdfs, Tests >Reporter: Mark Miller >Assignee: Kevin Risden >Priority: Major > Attachments: SOLR-5007.patch > > -- 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] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16760202#comment-16760202 ] Kevin Risden commented on SOLR-5007: I think I found the underlying issue. FileContext.getFileContext doesn't close the underlying filesystem that gets created. There is a rename method now on the Filesystem that gets closed correctly. I am trying to track down when the rename method was added to Filesystem class. I need to run all the HDFS related tests to make sure nothing is broken but this fixed the IPC thread issue with a few of the tests (TestRecoveryHdfs, HdfsDirectoryFactoryTest, HdfsDirectoryTest). > TestRecoveryHdfs seems to be leaking a thread occasionally that ends up > failing a completely different test. > > > Key: SOLR-5007 > URL: https://issues.apache.org/jira/browse/SOLR-5007 > Project: Solr > Issue Type: Test > Components: Hadoop Integration, hdfs, Tests >Reporter: Mark Miller >Assignee: Kevin Risden >Priority: Major > Attachments: SOLR-5007.patch > > -- 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] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16760043#comment-16760043 ] Kevin Risden commented on SOLR-5007: [~hossman] - sorry you are right. I didn't see the commits get moved around. I knew BadHdfsThreadsFilter existed but didn't grep for SOLR-5007. I can take a look and see if this is actually resolved with SOLR-5007 "fixes/workarounds" removed. > TestRecoveryHdfs seems to be leaking a thread occasionally that ends up > failing a completely different test. > > > Key: SOLR-5007 > URL: https://issues.apache.org/jira/browse/SOLR-5007 > Project: Solr > Issue Type: Test > Components: Hadoop Integration, hdfs, Tests >Reporter: Mark Miller >Assignee: Kevin Risden >Priority: Major > -- 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] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16759196#comment-16759196 ] Kevin Risden commented on SOLR-5007: Its been 3+ years since last comment. Not sure if these threads are still leaking. Hadoop is now upgraded to Hadoop 3 from SOLR-9515. SOLR-7289 may have also fixed this issue or at least limited it to known failing threads. Planning to close this since not much going on here to work from. > TestRecoveryHdfs seems to be leaking a thread occasionally that ends up > failing a completely different test. > > > Key: SOLR-5007 > URL: https://issues.apache.org/jira/browse/SOLR-5007 > Project: Solr > Issue Type: Test >Reporter: Mark Miller >Assignee: Mark Miller >Priority: Major > -- 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] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14377159#comment-14377159 ] Mark Miller commented on SOLR-5007: --- bq. I never got to this. I'll work on that today. I've spun this off to SOLR-7289. TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test. Key: SOLR-5007 URL: https://issues.apache.org/jira/browse/SOLR-5007 Project: Solr Issue Type: Test Reporter: Mark Miller Assignee: Mark Miller -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14375133#comment-14375133 ] Mark Miller commented on SOLR-5007: --- I've tried for 2.6.0 as well and this thread can still leak. I think in the simple case it is fixed, but if you trigger lease recover it still leaks along with a couple threads noted in related JIRA issues. TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test. Key: SOLR-5007 URL: https://issues.apache.org/jira/browse/SOLR-5007 Project: Solr Issue Type: Test Reporter: Mark Miller Assignee: Mark Miller -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14372998#comment-14372998 ] Mark Miller commented on SOLR-5007: --- Hmm...it was fixed for 2.3.0 and we are now on 2.3.0, but I still see this as a problem. I'll try to investigate sometime soon. TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test. Key: SOLR-5007 URL: https://issues.apache.org/jira/browse/SOLR-5007 Project: Solr Issue Type: Test Reporter: Mark Miller Assignee: Mark Miller -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14373001#comment-14373001 ] Mark Miller commented on SOLR-5007: --- bq. Yep, it's better to ignore those known threads than ignore everything. I never got to this. I'll work on that today. TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test. Key: SOLR-5007 URL: https://issues.apache.org/jira/browse/SOLR-5007 Project: Solr Issue Type: Test Reporter: Mark Miller Assignee: Mark Miller -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13708027#comment-13708027 ] Mark Miller commented on SOLR-5007: --- HADOOP-9703 has been resolved, targeted for 2.2.0 TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test. Key: SOLR-5007 URL: https://issues.apache.org/jira/browse/SOLR-5007 Project: Solr Issue Type: Test Reporter: Mark Miller Assignee: Mark Miller -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13703321#comment-13703321 ] Mark Miller commented on SOLR-5007: --- HADOOP-9703 seems to be making some progress. bq. Also: we could add those IPC threads to the thread leak filters if they're harmless instead of doing Scope.NONE. That's probably preferable for now then - they should be harmless and they affect all the hdfs tests - having Scope.NONE makes it easy to introduce new leaks without noticing. TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test. Key: SOLR-5007 URL: https://issues.apache.org/jira/browse/SOLR-5007 Project: Solr Issue Type: Test Reporter: Mark Miller Assignee: Mark Miller -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13703371#comment-13703371 ] Dawid Weiss commented on SOLR-5007: --- Yep, it's better to ignore those known threads than ignore everything. TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test. Key: SOLR-5007 URL: https://issues.apache.org/jira/browse/SOLR-5007 Project: Solr Issue Type: Test Reporter: Mark Miller Assignee: Mark Miller -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13700479#comment-13700479 ] Dawid Weiss commented on SOLR-5007: --- Yeah, HADOOP-9703 looks like the core of the problem, thanks Mark. I still need to figure out how the thread group propagates -- I have it on my list. TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test. Key: SOLR-5007 URL: https://issues.apache.org/jira/browse/SOLR-5007 Project: Solr Issue Type: Test Reporter: Mark Miller Assignee: Mark Miller -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13700530#comment-13700530 ] Dawid Weiss commented on SOLR-5007: --- Ok, I think I know. Whenever you're using this: {code} @ThreadLeakScope(Scope.NONE) // hdfs mini cluster currently leaks threads {code} it means any threads this test leaves behind will be a problem to debug. And any threads they themselves create will cause a thread leak later. So it's not a bug in the test framework. This annotation is present in a number of classes; it'd be best to get rid of it as soon as possible... TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test. Key: SOLR-5007 URL: https://issues.apache.org/jira/browse/SOLR-5007 Project: Solr Issue Type: Test Reporter: Mark Miller Assignee: Mark Miller -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13700531#comment-13700531 ] Dawid Weiss commented on SOLR-5007: --- Also: we could add those IPC threads to the thread leak filters if they're harmless instead of doing Scope.NONE. TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test. Key: SOLR-5007 URL: https://issues.apache.org/jira/browse/SOLR-5007 Project: Solr Issue Type: Test Reporter: Mark Miller Assignee: Mark Miller -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13700102#comment-13700102 ] Dawid Weiss commented on SOLR-5007: --- Hmm... within the same suite? If it's a different suite then this shouldn't be possible. Let me know, Mark. TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test. Key: SOLR-5007 URL: https://issues.apache.org/jira/browse/SOLR-5007 Project: Solr Issue Type: Test Reporter: Mark Miller Assignee: Mark Miller -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13700117#comment-13700117 ] Mark Miller commented on SOLR-5007: --- An example of what I am seeing: http://jenkins.thetaphi.de/job/Lucene-Solr-trunk-Linux/6421/ TestRandomDVFaceting fails with: {noformat} Error Message: 1 thread leaked from SUITE scope at org.apache.solr.TestRandomDVFaceting: 1) Thread[id=4381, name=IPC Parameter Sending Thread #2, state=TIMED_WAITING, group=TGRP-TestRecoveryHdfs] at {noformat} I assume it's due to TestRecoveryHdfs because of group=TGRP-TestRecoveryHdfs TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test. Key: SOLR-5007 URL: https://issues.apache.org/jira/browse/SOLR-5007 Project: Solr Issue Type: Test Reporter: Mark Miller Assignee: Mark Miller -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13700348#comment-13700348 ] Dawid Weiss commented on SOLR-5007: --- This: {code} Thread[id=4381, name=IPC Parameter Sending Thread #2, state=TIMED_WAITING, group=TGRP-TestRecoveryHdfs] {code} means the thread is indeed part of the thread group formed for TestRecoveryHdfs. There is a catch though -- the thread that leaked from this test couldn't exist back then (because it'd be marked as existing at the start of the test). My suspicion is that what happens is an IPC thread is created in some test and then this thread spawns more threads on-demand or something. So the thread group of that parent, persistent thread is inherited by the children who are then detected as rogue threads. I cannot explain why that parent thread isn't causing a thread leak error -- it'd have to be in the set of ignored threads (but its children are not). If anything comes to your mind, shoot. I'll try to investigate tomorrow. TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test. Key: SOLR-5007 URL: https://issues.apache.org/jira/browse/SOLR-5007 Project: Solr Issue Type: Test Reporter: Mark Miller Assignee: Mark Miller -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13700411#comment-13700411 ] ASF subversion and git services commented on SOLR-5007: --- Commit 1499894 from [~markrmil...@gmail.com] [ https://svn.apache.org/r1499894 ] SOLR-5007: clean up resource management TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test. Key: SOLR-5007 URL: https://issues.apache.org/jira/browse/SOLR-5007 Project: Solr Issue Type: Test Reporter: Mark Miller Assignee: Mark Miller -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13700412#comment-13700412 ] ASF subversion and git services commented on SOLR-5007: --- Commit 1499895 from [~markrmil...@gmail.com] [ https://svn.apache.org/r1499895 ] SOLR-5007: clean up resource management TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test. Key: SOLR-5007 URL: https://issues.apache.org/jira/browse/SOLR-5007 Project: Solr Issue Type: Test Reporter: Mark Miller Assignee: Mark Miller -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] [Commented] (SOLR-5007) TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test.
[ https://issues.apache.org/jira/browse/SOLR-5007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13700413#comment-13700413 ] Mark Miller commented on SOLR-5007: --- I tried filing HADOOP-9703 TestRecoveryHdfs seems to be leaking a thread occasionally that ends up failing a completely different test. Key: SOLR-5007 URL: https://issues.apache.org/jira/browse/SOLR-5007 Project: Solr Issue Type: Test Reporter: Mark Miller Assignee: Mark Miller -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org