[
https://issues.apache.org/jira/browse/SOLR-2715?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Hoss Man updated SOLR-2715:
---------------------------
Attachment: SOLR-2715.patch
Since i got sucked into looking at the JMX tests yesterday, i spent 24+ hours
digging into this too see if i could resolve it as well.
After re-working hte test to do an apples to apples test instead of comparing
JMX with the infoRegistery (to decouple from the problems mentioned in
SOLR-5095) i was still seeing extremely sporadic failures (off by one) from the
this reload test.
at first i thought perhaps this was a timing issue related to core reload
blocking and updated the test accordingly -- but that didn't seem to have any
effect. Then i realized that there were some overlapping searchers happening
in the test (not sure why exactly: no commits happening) and i realized that
since we register searchers with multiple names ("currentSearcher" for general
monitoring and "Searcher@<id>" for specific checks) that could cause a problem
during the overlap period. When i excluded them from the count, i couldn't
reproduce those failures anymore, but i did start seeing
InstanceNotFoundException exceptions when looking at caches -- which also makes
sense in an onDeck searcher situation since it also re-registers the caches.
So i *think* this patch should get things to the point where the meat of the
goal (ensuring everything is reregistered after core reload) is still testted,
but we shoulnd't get sporadic failures.
would appreciate if folks could let me know if they can get any failures from
this patch...
{noformat}
ant test -Dtests.failfast=yes -Dtests.iters=10000 -Dtestcase=TestJmxIntegration
-Dtests.method=testJmxOnCoreReload\*
{noformat}
> TestJMXSolrIntegration fails
> ----------------------------
>
> Key: SOLR-2715
> URL: https://issues.apache.org/jira/browse/SOLR-2715
> Project: Solr
> Issue Type: Bug
> Affects Versions: 3.4
> Reporter: Robert Muir
> Attachments: SOLR-2715.patch
>
>
> Running the tests, this test fails (in a non-reproducible way). There might
> be some sort of timing issue here.
--
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
