[
https://issues.apache.org/jira/browse/SOLR-7603?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14565617#comment-14565617
]
Hoss Man commented on SOLR-7603:
--------------------------------
TL;DR: still no clue, but added more logging/assert details to test
i've been beating my head against this and still can't make heads or tells of
this failure -- the best guess i've got is that the logic for "pruning" the
distributed chain down (but including any "RunAllways" processors) is actually
working fine, but perhaps there is some failure in the initial construction of
the chain in the first place? (SOLR-6892 recently modified the way the chains
are initialized)
So i've added a hack to increase the log level for the duration of the test, as
well as some more asserts regarding the state of the chain, and simplified the
logic around how we assert properties of the distributed chain so it's a bit
more straight forward an we can include a list of every proc in every assert.
> Scary non reproducible failure from
> UpdateRequestProcessorFactoryTest.testUpdateDistribChainSkipping
> ----------------------------------------------------------------------------------------------------
>
> Key: SOLR-7603
> URL: https://issues.apache.org/jira/browse/SOLR-7603
> Project: Solr
> Issue Type: Bug
> Reporter: Hoss Man
> Attachments: SOLR-7603.consoleText.txt
>
>
> jenkins nightly hit a very inexplicable error today...
> {noformat}
> Build: https://builds.apache.org/job/Lucene-Solr-NightlyTests-5.x/860/
> At revision 1682097
> Updating http://svn.apache.org/repos/asf/lucene/dev/branches/branch_5x at
> revision '2015-05-27T14:50:50.016 -0400'
> [java-info] java version "1.7.0_72"
> [java-info] Java(TM) SE Runtime Environment (1.7.0_72-b14, Oracle Corporation)
> [java-info] Java HotSpot(TM) 64-Bit Server VM (24.72-b04, Oracle Corporation)
> {noformat}
> {noformat}
> [junit4] 2> NOTE: reproduce with: ant test
> -Dtestcase=UpdateRequestProcessorFactoryTest
> -Dtests.method=testUpdateDistribChainSkipping -Dtests.seed=4ECABCCFD159BE21
> -Dtests.multiplier=2
> -Dtests.nightly=true -Dtests.slow=true
> -Dtests.linedocsfile=/x1/jenkins/lucene-data/enwiki.random.lines.txt
> -Dtests.locale=mt_MT -Dtests.timezone=Etc/GMT0 -Dtests.asserts=true
> -Dtests.file.encoding=ISO-8859-1
> [junit4] FAILURE 0.01s J0 |
> UpdateRequestProcessorFactoryTest.testUpdateDistribChainSkipping <<<
> [junit4] > Throwable #1: java.lang.AssertionError
> [junit4] > at
> __randomizedtesting.SeedInfo.seed([4ECABCCFD159BE21:3F2E4219A7B299ED]:0)
> [junit4] > at
> org.apache.solr.update.processor.UpdateRequestProcessorFactoryTest.testUpdateDistribChainSkipping(UpdateRequestProcessorFactoryTest.java:111)
> [junit4] > at java.lang.Thread.run(Thread.java:745)
> {noformat}
> ...the line in question is asserting that when executing a distributed update
> (ie: forwarded from another node), the LogUpdateProcessor is still part of
> the chain because it's got got the "RunAlways" annotation indicating it
> should always be included in the chain (everything before hte
> DistribUpdateProcessor is normally)
> There's really no explanation for why the LogUpdateProcessor wouldn't be
> found other then a code bug -- but in that case why doesn't the seed
> reproduce reliably?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
