[jboss-jira] [JBoss JIRA] (WFLY-2838) Revisit default mod_cluster metric for WildFly 8

[Jason Greene (JIRA)]

    [ https://issues.jboss.org/browse/WFLY-2838?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12940622#comment-12940622 ] 

Jason Greene commented on WFLY-2838:
------------------------------------

I think its hard to determine based on number of requests because it all depends on how heavy the user code is in the context of a request.

CPU time is a pretty good value to use IMO. We could still have a watermark that the user could override. Alternatively it could get fancier and calculate the amount of CPU time used by the I/O threads and the worker pool saturation. That would require deeper OS integration and more work than we have time left.
                
> Revisit default mod_cluster metric for WildFly 8
> ------------------------------------------------
>
>                 Key: WFLY-2838
>                 URL: https://issues.jboss.org/browse/WFLY-2838
>             Project: WildFly
>          Issue Type: Feature Request
>      Security Level: Public(Everyone can see) 
>          Components: Clustering
>    Affects Versions: 8.0.0.CR1
>            Reporter: Radoslav Husar
>            Assignee: Radoslav Husar
>            Priority: Blocker
>             Fix For: 8.0.0.Final
>
>
> Since the threading aspects of Undertow in WildFly are completely different from JBoss Web's in the previous versions of AS, we should reconsider the default metric for WildFly.
> Currently, busyness requires explicit capacity to be useful which would require further tweaking by users.
> Metric "cpu" is also a good candidate since it doesn't need explicit capacity.. 

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira