[
https://issues.apache.org/jira/browse/SOLR-7836?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14661898#comment-14661898
]
Erick Erickson commented on SOLR-7836:
--------------------------------------
Well, something that sits in a while loop waiting for another thread to change
a variable before doing anything seems like it's harder to get right than some
kind of notification. As I'm finding. But I'm terrified of changing the total
approach.
I need to dig some more before posting stack traces, plus it's the trace with
this patch applied (and maybe changed)....
Mostly bringing it up to see if someone looks at the loops and says "Yuck, how
did we do _THAT_?" before thinking about replacing the mechanism. You've shot
that hope in the head though ;). And I'm assuming that part of the complexity
here is that we want the index writer to be held by a thread as briefly as
possible so as to not introduce bottlenecks.
Anyway, I'll dig some more after I take care of morning tasks.
Thanks!
> Possible deadlock when closing refcounted index writers.
> --------------------------------------------------------
>
> Key: SOLR-7836
> URL: https://issues.apache.org/jira/browse/SOLR-7836
> Project: Solr
> Issue Type: Bug
> Reporter: Erick Erickson
> Assignee: Erick Erickson
> Attachments: SOLR-7836.patch
>
>
> Preliminary patch for what looks like a possible race condition between
> writerFree and pauseWriter in DefaultSorlCoreState.
> Looking for comments and/or why I'm completely missing the boat.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
