Mailing-List: contact activemq-dev-help@geronimo.apache.org; run by ezmlm
Precedence: bulk
Reply-To: activemq-dev@geronimo.apache.org
Message-ID: <8323601.1164438722615.JavaMail.root@brutus>
Date: Fri, 24 Nov 2006 23:12:02 -0800 (PST)
From: "Rob Davies (JIRA)" <jira@apache.org>
To: activemq-dev@geronimo.apache.org
Subject: [jira] Resolved: (AMQ-791) support spool to disk for non-persistent
 topic consumers
In-Reply-To: <27853751.1151951631479.JavaMail.jira@brutus>
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

     [ https://issues.apache.org/activemq/browse/AMQ-791?page=all ]

Rob Davies resolved AMQ-791.
----------------------------

    Resolution: Fixed

This is SVN revision 479094

Use in mmeory list in FilePendingMessageCursor - until memory limit reached - then use disk.
USe FilePendingMessageCursor in TopicSubscription instead of LinkedList

> support spool to disk for non-persistent topic consumers
> --------------------------------------------------------
>
>                 Key: AMQ-791
>                 URL: https://issues.apache.org/activemq/browse/AMQ-791
>             Project: ActiveMQ
>          Issue Type: New Feature
>          Components: Broker
>            Reporter: james strachan
>         Assigned To: Rob Davies
>             Fix For: 4.2.0
>
>
> Rather than just blocking when RAM is full we could have a high-water mark where we start spooling messages to disk if there is not sufficient RAM to hold the messages.
> The good thing about this approch is that it avoids blocking the producers when RAM is full; the downside is that once spooling starts, the producer will be slowed down to the speed of the disk spooling (as due to RAM exhaustion under steady state, the producer will have to wait for the message to be spooled to disk so that it can evict it from RAM so that it can send the next message).
> Though the journal is quite fast so the slow down shouldn't be too many orders of magnitude (and is better than making things appear to 'lock up' while we wait for the slowest consumer to acknowledge more messages).

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: https://issues.apache.org/activemq/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira