[
https://issues.apache.org/jira/browse/SOLR-9363?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15403074#comment-15403074
]
Hoss Man commented on SOLR-9363:
--------------------------------
I should have mentioned earlier, but the particular assertion that fails is
before the test even gets started on anything remotely interesting related to
atomic updates -- it's a beginning of the test sanity check that the index is
starting with the expected number of documents.
After getting hung up on a small red herring for a bit (git bisect identified a
commit related to LUCENE-7390 -- which has to do with IndexWriter buffering --
as being where this seed started fail, but it turns out RandomCodec was
modified in that commit to consume more entropy, which makes it meaninless if
this seed passed prior to that commit) I've been wading through things and i
think ultimatley i've identified 2 unrelated problems:
# TestInjection's use of Random doesn't create reproducible sequences of events
when running individual test methods vs an entire class
# replicas which are in the middle of recovery will ignore explicit commits
(ie: {{Ignoring commit while not ACTIVE - state: BUFFERING replay: false}}) but
will aparently still participate in search requests.
...neither of which really has anything to do with atomic updates.
I'll file more detailed bugs on both of these issues, in the meantime i'm
pretty sure that in this test i can work around the combination of #1 + #2 ...
details once i do some experimenting.
> TestStressCloudBlindAtomicUpdates.test_dv_stored() failure
> ----------------------------------------------------------
>
> Key: SOLR-9363
> URL: https://issues.apache.org/jira/browse/SOLR-9363
> Project: Solr
> Issue Type: Bug
> Security Level: Public(Default Security Level. Issues are Public)
> Reporter: Steve Rowe
> Attachments: Lucene-Solr-Nightly-master.271.log.gz
>
>
> My Jenkins found a reproducing seed - though note that in order to reproduce
> I had to a) run the test at the same SHA as the one listed, and b) leave off
> {{-Dtests.method=test_dv_stored}}; under those conditions it failed both
> times I ran it:
> {noformat}
> Checking out Revision 53a34b312e78ce6f56c0bb41304ac834b28b9534
> (refs/remotes/origin/master)
> [...]
> [junit4] 2> NOTE: reproduce with: ant test
> -Dtestcase=TestStressCloudBlindAtomicUpdates -Dtests.method=test_dv_stored
> -Dtests.seed=60C9AAFC94FD8921 -Dtests.multiplier=2 -Dtests.nightly=true
> -Dtests.slow=true
> -Dtests.linedocsfile=/home/jenkins/lucene-data/enwiki.random.lines.txt
> -Dtests.locale=id -Dtests.timezone=America/Puerto_Rico -Dtests.asserts=true
> -Dtests.file.encoding=UTF-8
> [junit4] FAILURE 9.09s J0 |
> TestStressCloudBlindAtomicUpdates.test_dv_stored <<<
> [junit4] > Throwable #1: java.lang.AssertionError: expected:<332> but
> was:<205>
> [junit4] > at
> __randomizedtesting.SeedInfo.seed([60C9AAFC94FD8921:F2F5CB7828BAB4F9]:0)
> [junit4] > at
> org.apache.solr.cloud.TestStressCloudBlindAtomicUpdates.checkField(TestStressCloudBlindAtomicUpdates.java:253)
> [junit4] > at
> org.apache.solr.cloud.TestStressCloudBlindAtomicUpdates.test_dv_stored(TestStressCloudBlindAtomicUpdates.java:192)
> [junit4] > at java.lang.Thread.run(Thread.java:745)
> {noformat}
--
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
