activemq-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Miroslav Novak (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ARTEMIS-541) Optimize live/backup replication after failback
Date Thu, 26 May 2016 15:02:15 GMT

    [ https://issues.apache.org/jira/browse/ARTEMIS-541?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15302193#comment-15302193
] 

Miroslav Novak commented on ARTEMIS-541:
----------------------------------------

I understand the reasons. Idea was to make something like index in the journal. It would be
sequential and would just grow. Every append to journal would have such sequential number.
Then if live would start, it would send its last index to backup. Backup would send everything
what is greater than the index.

> Optimize live/backup replication after failback
> -----------------------------------------------
>
>                 Key: ARTEMIS-541
>                 URL: https://issues.apache.org/jira/browse/ARTEMIS-541
>             Project: ActiveMQ Artemis
>          Issue Type: New Feature
>          Components: Broker
>    Affects Versions: 1.3.0
>            Reporter: Miroslav Novak
>
> Currently if there is master/slave pair configured using replicated journal then after
each failback when master is starting, it copies the whole journal directory from slave.
> This seems to ineffective as master might have significant part of backups journal before
it failed. 
> If only differences between journals would be transfered then it would be effective especially
in case when failback is completed and master starts to synchronize the whole journal back
to slave. It copies the whole journal to backup again which is almost the same. This would
greatly speed up synchronization of slave and safe network bandwidth.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message