[
https://issues.apache.org/jira/browse/SOLR-14651?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17214949#comment-17214949
]
David Smiley commented on SOLR-14651:
-------------------------------------
I'm looking back at this. When enable="false", it's disabled/disfunctional but
still exists, and thus you can try to communicate with it but you'll end up
getting an NPE. IMO, I should have gone further – enable=false should result
in no instantiation of the metrics history handler at all. Similar to Mark
Miller's branch. That said, the documentation on "enable" at the above link
says:
{quote}If this is {{false}} then metrics history is not collected but can still
be retrieved from existing databases.
{quote}
[~ab] is that really intentional? If so, then my change committed here (this
issue) breaks that. Should I simply update the docs or...?
> Metrics history could disable itself better
> -------------------------------------------
>
> Key: SOLR-14651
> URL: https://issues.apache.org/jira/browse/SOLR-14651
> Project: Solr
> Issue Type: Improvement
> Components: metrics
> Reporter: David Smiley
> Assignee: David Smiley
> Priority: Minor
> Fix For: 8.7
>
> Time Spent: 50m
> Remaining Estimate: 0h
>
> The MetricsHistoryHandler should disable itself better than it does. It will
> create a SolrRrdBackendFactory (which has an ExecutorService / thread pool)
> when it should not when the plugin is disabled. Furthermore, I think this
> should be disabled by default in tests, as it's excessive machinery that
> don't matter to 99.9% of tests. This will help our tests run more
> efficiently.
> While working on other issues, I also occasionally saw a thread leak failure
> with the RRD being the leaked threads. I see that in "close", we don't
> _wait_ for the executor to actually shut down but it should.
> CC [~ab]
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]