Mailing-List: contact dev-help@couchdb.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@couchdb.apache.org
Date: Wed, 2 Feb 2011 18:53:29 +0000 (UTC)
From: "Filipe Manana (JIRA)" <jira@apache.org>
To: dev@couchdb.apache.org
Message-ID: <4334568.5663.1296672809011.JavaMail.tomcat@hel.zones.apache.org>
In-Reply-To: 
 <620453889.4421.1296621749066.JavaMail.tomcat@hel.zones.apache.org>
Subject: [jira] Commented: (COUCHDB-1054) Better couch_log performance
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/COUCHDB-1054?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12989743#comment-12989743 ] 

Filipe Manana commented on COUCHDB-1054:
----------------------------------------

Yes, the addition of the timestamp is not externalized to avoid having strange logs, with later timestamps coming before earlier ones. If externalized, it would likely confuse many users. However I don't mind at all about it.

Is everyone ok with this?


At some point I had the message converted to a binary with an iolist_to_binary call, it's certainly difficult to measure the impact, but I'm not sure what is more expensive: converting to a binary or copying an iolist.
I didn't know about io:put_chars, but it seems to accept binaries and iolists besides charlists (the docs only mention charlists). I'll definitely use it.


> Better couch_log performance
> ----------------------------
>
>                 Key: COUCHDB-1054
>                 URL: https://issues.apache.org/jira/browse/COUCHDB-1054
>             Project: CouchDB
>          Issue Type: Improvement
>            Reporter: Filipe Manana
>            Assignee: Filipe Manana
>         Attachments: COUCHDB-1054-2.patch, COUCHDB-1054.patch
>
>
> Building the messages to write to the console and the log file can be done outside the couch_log gen_event. This significantly increases the parallelism when many processes log messages. The following relaximation test graph shows a throughput increase using the attached patch against current trunk:
> http://graphs.mikeal.couchone.com/#/graph/0379dbdaef29b1c0fbf03421540243f7

-- 
This message is automatically generated by JIRA.
-
For more information on JIRA, see: http://www.atlassian.com/software/jira