[
https://issues.apache.org/jira/browse/GEODE-5315?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16508959#comment-16508959
]
Sean Goller commented on GEODE-5315:
------------------------------------
We attached a class that exhibits the behavior.
> LuceneService.waitForFlushed does not wait if members have been restarted
> -------------------------------------------------------------------------
>
> Key: GEODE-5315
> URL: https://issues.apache.org/jira/browse/GEODE-5315
> Project: Geode
> Issue Type: Bug
> Components: lucene
> Reporter: Sean Goller
> Priority: Major
> Attachments: GeodeLuceneWaitForFlushedTest.java
>
>
> We wrote a test that
> * puts 10K entries,
> * pauses the async event queue
> * removes the same entries minus 1
> * Closes the cache in the data stores
> * Recreates the cache
> * calls LuceneService.waitForFlush
> * Does a query to LuceneService ensuring only 1 item is in the cache
> This test fails consistently, even though we would expect waitForFlush not to
> return until the queue has been fully processed. Running the same test
> without closing/recreating the cache succeeds consistently, so there seems to
> be an interaction problem between some combination of the queue, cache
> lifecycle and waitForFlush.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
