[
https://issues.apache.org/jira/browse/SOLR-13454?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16838110#comment-16838110
]
Erick Erickson commented on SOLR-13454:
---------------------------------------
I just noticed a test failure in ReindexCollectionTest, but I don't see how it
could have been caused by my test experiments. That said, in looking at my test
experimnts I realized I put them a line too low, fixing.
It's strange that the NPE is happening in this block of code:
{code:java}
cloudManager.getClusterStateProvider().getClusterState().forEachCollection(coll
-> {
assertFalse(coll.getName() + " still exists",
coll.getName().startsWith(ReindexCollectionCmd.TARGET_COL_PREFIX));
assertFalse(coll.getName() + " still exists",
coll.getName().startsWith(ReindexCollectionCmd.CHK_COL_PREFIX));
});
{code}
> Investigate ReindexCollectionTest failures
> ------------------------------------------
>
> Key: SOLR-13454
> URL: https://issues.apache.org/jira/browse/SOLR-13454
> Project: Solr
> Issue Type: Improvement
> Security Level: Public(Default Security Level. Issues are Public)
> Reporter: Erick Erickson
> Assignee: Erick Erickson
> Priority: Minor
>
> This _looks_ like it might be another example of commits not quite happening
> correctly, see
> SOLR-11035. Problem is I can’t get it to fail locally after 2,000 iterations.
> So I’m going to add a bit to the bandaid to allow tests to conditionally fail
> if the bandaid would have made it pass. That way we can positively detect
> that the bandaid is indeed the case rather than change code and hope.
> This _shouldn’t_ add any noise to the Jenkins lists, as the test won’t fail
> in cases where it didn’t before.
> In case people wonder what the heck I’m doing.
> BTW, if we ever really understand/fix the underlying cause, we should make
> the bandaid code fail and see, then remove it if so.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
