Mailing-List: contact solr-dev-help@lucene.apache.org; run by ezmlm
Precedence: bulk
Reply-To: solr-dev@lucene.apache.org
Message-ID: <676199520.15291267308905825.JavaMail.jira@brutus.apache.org>
Date: Sat, 27 Feb 2010 22:15:05 +0000 (UTC)
From: "Bill Bell (JIRA)" <jira@apache.org>
To: solr-dev@lucene.apache.org
Subject: [jira] Commented: (SOLR-1775) Replication of 300MB stops indexing
 for 5 seconds when syncing
In-Reply-To: <1362534500.292011266297868054.JavaMail.jira@brutus.apache.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/SOLR-1775?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12839318#action_12839318 ] 

Bill Bell commented on SOLR-1775:
---------------------------------

I agree it is i/o bound. But when we sync using Java replication, the slave STOPS taking requests for about 5 seconds. I.E.

1. The sync begins - initiated by the slave (the files are almost 1GB)
2. The slave is still taking requests
3. The slave completes the Sync
4. The requests to the slave STOPS for 5 seconds.
5. The slave continues taking requests

I think the copy from one dir to another of a 1GB file is slowing down the machine - the i/o waits are like 50%. Is there a way to reduce the impact of the copy and switchover?


> Replication of 300MB stops indexing for 5 seconds when syncing
> --------------------------------------------------------------
>
>                 Key: SOLR-1775
>                 URL: https://issues.apache.org/jira/browse/SOLR-1775
>             Project: Solr
>          Issue Type: Bug
>          Components: replication (java)
>    Affects Versions: 1.4
>         Environment: Centos 5.3
>            Reporter: Bill Bell
>
> When using Java replication in v1.4 and doing a sync from master to slave, the slave delays for about 5-10 seconds. When using rsync this does not occur.
> Is there a way to thread better or lower the priority to not impact queries when it is bringing over the index files from the master? Maybe a separate process?

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.