[
https://issues.apache.org/jira/browse/SOLR-7836?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14663038#comment-14663038
]
Erick Erickson commented on SOLR-7836:
--------------------------------------
Looking a bit more, all the ulog.(some operation) commands have this at the
top:
{code}
synchronized (this) {
{code}
so they're serialized. I don't like the fact that lots of these are called from
within
{code}
synchronized (solrCoreState.getUpdateLock()) {
{code}
as that seems like a good place for a deadlock if it ever happens that the ulog
operations want to get that same lock, but I'm not about to change that now by,
say, moving all the ulog operations out of the synchronized blocks since I
don't know why they were put in the first place, we can make that a separate
JIRA if necessary/desirable. I'm comfortable enough with the one I moved out to
go ahead with this commit.
So I'll clean up the patch and commit unless there are objections
today/tomorrow.
> 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, 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]
