[
https://issues.apache.org/jira/browse/RANGER-1429?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16970869#comment-16970869
]
Kevin Risden commented on RANGER-1429:
--------------------------------------
It looks like this still patch is still applicable to current Solr configs for
Ranger. Spellcheck in a thread used to be an issue (not sure if it was fixed in
Solr itself). Either way, if Ranger doesn't need spellcheck seems like an easy
win.
[~aarrieta] have you run into this as well? Specifically around infra-solr?
This seems like a reasonable suggestion to prevent spell check for the
collection.
[~vel] not sure if spell check is used in Ranger anywhere. I don't have any
examples off the top of my head.
> Ranger Audit solrconfig.xml optimizations
> -----------------------------------------
>
> Key: RANGER-1429
> URL: https://issues.apache.org/jira/browse/RANGER-1429
> Project: Ranger
> Issue Type: Bug
> Components: audit
> Affects Versions: 0.7.0
> Reporter: Greg Senia
> Priority: Major
> Attachments: RANGER-1429.patch
>
>
> When using SolrCloud and having a single shard single replica with a large
> number of audit records 20GB for 2 weeks. It takes Solr a very long time to
> open up the replica for work. After doing some research I determine via
> thread dumps that on the first request to the Solr Collection it kicks off a
> Suggestion thread and a SpellCheck thread. I propose removing these as it
> does not seem to impact RangerAudit and solves the issue of Solr
> ranger_audits collection being unavailable after solr startup.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
