[
https://issues.apache.org/jira/browse/SOLR-13530?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16859608#comment-16859608
]
Anshum Gupta commented on SOLR-13530:
-------------------------------------
In addition to the comment in the parent Jira :
https://issues.apache.org/jira/browse/SOLR-13527?focusedCommentId=16859595&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#comment-16859595
This processor will be configured by setting the following params:
{code:xml}
<processor class="CoreSizeCheckerUpdateProcessorFactory">
<str name="coresizechecker.enabled">true</str>
<str name="coresizechecker.dryRun">false</str>
<str name="coresizechecker.blockLevel">collection</str>
<str name="coresizechecker.blockAddIndexSizeInBytes">5k</str>
<str name="coresizechecker.unBlockAddIndexSizeInBytes">1k</str>
</processor>{code}
*blockAddIndexSizeInBytes* - Size of the core, beyond which the update requests
would start getting rejected
*unBlockAddIndexSizeInBytes* - Size of the core below which, the update
requests would get unblocked if already blocked.
This can currently be configured to run at collection or shard level i.e. if
the core size grows beyond a specific value, you may want to either disable
updates for that specific core or for the entire collection. The recommended
practice in my opinion would be to just disable it for the collection level to
not cause confusion, unless you manage your routing in a multi-shard setup.
> Control requests to Solr based on the core size
> -----------------------------------------------
>
> Key: SOLR-13530
> URL: https://issues.apache.org/jira/browse/SOLR-13530
> Project: Solr
> Issue Type: Sub-task
> Reporter: Anshum Gupta
> Assignee: Anshum Gupta
> Priority: Major
>
> Large cores are almost always problematic. They show up as increased
> replication times, recovery times, and much more. They also often are
> attempted to be handled by using a larger heap, which is also almost always
> an even bigger problem.
> We should be able to reject requests or at least log/notify when Solr core
> grows beyond a specific threshold.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
