You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@lucene.apache.org by "Cassandra Targett (JIRA)" <ji...@apache.org> on 2016/11/30 21:55:58 UTC
[jira] [Closed] (SOLR-4924) indices getting out of sync with
SolrCloud
[ https://issues.apache.org/jira/browse/SOLR-4924?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Cassandra Targett closed SOLR-4924.
-----------------------------------
Resolution: Won't Fix
As Mark notes, a lot has changed here to make this less likely to still be a problem (or manifests very differently than described here).
> indices getting out of sync with SolrCloud
> ------------------------------------------
>
> Key: SOLR-4924
> URL: https://issues.apache.org/jira/browse/SOLR-4924
> Project: Solr
> Issue Type: Bug
> Components: replication (java), SolrCloud
> Affects Versions: 4.2
> Environment: Linux 2.6.18-308.16.1.el5 #1 SMP Tue Oct 2 22:01:43 EDT 2012 x86_64 x86_64 x86_64 GNU/Linux
> CentOS release 5.8 (Final)
> Solr 4.2.1
> Reporter: Ricardo Merizalde
>
> We are experiencing an issue in our production servers where the indices get out of sync. Customers will see different results/result sorting depending of the instance that serves the request.
> We currently have 2 instances with a single shard. This is our update handler configuration
> <updateHandler class="solr.DirectUpdateHandler2">
> <autoCommit>
> <!-- do a hard commit every 10 minutes -->
> <maxTime>600000</maxTime>
> <!-- or when there are more than 5000 pending documents -->
> <maxDocs>5000</maxDocs>
> <!-- save the changes to persistent storage but don't open a new search -->
> <openSearcher>false</openSearcher>
> </autoCommit>
> <autoSoftCommit>
> <!-- make changes visible every 5 seconds -->
> <maxTime>5000</maxTime>
> </autoSoftCommit>
> <updateLog>
> <str name="dir">${solr.data.dir:}</str>
> </updateLog>
> </updateHandler>
> When the indices get out of sync the follower replica ends up with a higher version than the master. Optimizing the leader or reloading the follower core doesn't not help. The only why to get the indices in sync is to restart the server.
> This is an state example of the leader:
> version: 1102541
> numDocs: 214007
> maxDoc: 370861
> deletedDocs: 156854
> While the follower core has the following state:
> version: 1109143
> numDocs: 213890
> maxDoc: 341585
> deletedDocs: 127695
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org