[
https://issues.apache.org/jira/browse/ARTEMIS-1452?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16194698#comment-16194698
]
ASF GitHub Bot commented on ARTEMIS-1452:
-----------------------------------------
Github user franz1981 commented on the issue:
https://github.com/apache/activemq-artemis/pull/1578
@clebertsuconic Thanks, now I see it!
Currently the MappedSequentialFileFactory doesn't extend the
AbstractSequentialFileFactory, hence it can't work for the MAPPED journal...
> Regarding MAPPED.. I didn't merge that one before because I thought we
were lacking tests... I had a test on my workspace for Unbuffered.. but I think
I missed it...
Yes, I don't know how the callbacks behave using unbuffered version, but if
it works for NIO for MAPPED it will work too..
> I think we should change the CLI to make timeout = 0 if sync = false (for
all the journals.. not just MAPPED). and add a text on the broker.xml saying we
recommend disabling if sync=false.
:100:
> Journal Configuration Improvements
> ----------------------------------
>
> Key: ARTEMIS-1452
> URL: https://issues.apache.org/jira/browse/ARTEMIS-1452
> Project: ActiveMQ Artemis
> Issue Type: Improvement
> Affects Versions: 2.3.0
> Reporter: clebert suconic
> Assignee: clebert suconic
> Fix For: 2.4.0
>
>
> When using more advanced systems, that are capable of doing a lot more queued
> IO, the self optimization of the journal is not really being optimal.
> Besides It would be nice to have some of the configurations more explicit on
> broker.xml
> And also, we should allow users to disable the TimedBuffering. By simply
> using 0 at the journal-buffer-timeout.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
