You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@lucene.apache.org by "Mike Drob (Jira)" <ji...@apache.org> on 2021/01/29 17:18:00 UTC
[jira] [Resolved] (SOLR-15085) EmbeddedSolrServer calls shutdown on
a provided CoreContainer
[ https://issues.apache.org/jira/browse/SOLR-15085?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Mike Drob resolved SOLR-15085.
------------------------------
Fix Version/s: 8.9
master (9.0)
Assignee: Mike Drob
Resolution: Fixed
Thanks for the contribution, Tim! I added a line to CHANGES and upgrade notes about this as well.
> EmbeddedSolrServer calls shutdown on a provided CoreContainer
> -------------------------------------------------------------
>
> Key: SOLR-15085
> URL: https://issues.apache.org/jira/browse/SOLR-15085
> Project: Solr
> Issue Type: Bug
> Security Level: Public(Default Security Level. Issues are Public)
> Components: Server, SolrJ
> Affects Versions: master (9.0)
> Reporter: Tim Owen
> Assignee: Mike Drob
> Priority: Minor
> Labels: pull-request-available
> Fix For: master (9.0), 8.9
>
> Time Spent: 1h 10m
> Remaining Estimate: 0h
>
> There are essentially 2 ways to create an EmbeddedSolrServer object, one by passing in a CoreContainer object, and the other way creates one internally on-the-fly. The current behaviour of the close method calls shutdown on the CoreContainer, regardless of where it came from.
> I believe this is not good behaviour for a class that doesn't control the lifecycle of the passed-in CoreContainer. In fact, there are 4 cases among the codebase where a subclass of EmbeddedSolrServer is created just to override this behaviour (with a comment saying it's unwanted).
> In my use-case I create EmbeddedSolrServer instances for cores as and when I need to work with them, but the CoreContainer exists for the duration. I don't want the whole container shut down when I'm done with just one of its cores. You can workaround it by just not calling close on the EmbeddedSolrServer object, but that's risky especially if you use a try-with-resources as close is called automatically then.
> Fix is to keep track of whether the CoreContainer was created internally or not, and only shut it down if internal. I will attach my patch PR.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@lucene.apache.org
For additional commands, e-mail: issues-help@lucene.apache.org