[
http://jira.codehaus.org/browse/CONTINUUM-2514?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Brett Porter updated CONTINUUM-2514:
------------------------------------
Fix Version/s: 1.4.1 (Beta)
> Prepare Build Queue slow down parallel built start
> --------------------------------------------------
>
> Key: CONTINUUM-2514
> URL: http://jira.codehaus.org/browse/CONTINUUM-2514
> Project: Continuum
> Issue Type: Bug
> Components: Parallel Builds
> Affects Versions: 1.3.6
> Reporter: Philippe Busque
> Fix For: 1.4.1 (Beta)
>
>
> When a scheduler or a build group trigger a series of compilation, every
> projects are enqueued in the Prepare Build Queue. This queue doesn't
> consider the project group of a project when iterating through it.
> As a result, newly added project, but independent, can wait a long time
> before being launched in a different built queue, as it has to wait for every
> entries from the previous groups to be processed.
> To speed up the launch of parallel builts, the Prepare Build Queue should
> alternate between Project Group when more than 1 group is present.
> Example:
> Set up 4 parallel built queues.
> Put 100 projects in Group A.
> Put 2 projects in Group B.
> Launch the built of Group A. All jobs are inserted in the Prepare Build Queue.
> Launch the built of Group B. The 2 jobs are inserted in the Prepare Build
> Queue.
> Since the projects of group B are independent to Group A, they should be
> added to the Built Queue immediately. However, they will only be added once
> the 100 projects from Group A has been analyzed.
> In our case, we have 6 queues and 40 groups. The average busy queue for the
> first half hour of compile is 4 queues as the prepare queue doesn't push new
> projects fast enough to fill the queues.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://jira.codehaus.org/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
