Re: Jenkins labels and large queue

["Doroszlai, Attila" <ad...@apache.org>]

Hi all,

I've turned on parallel builds for Ambari-Github-PullRequest-Builder.
I didn't see any problems due to this so far.  Most of the time builds
land on different nodes.  Even if they are executed on the same node
concurrently, separate workspace directories are used.  But please let
me know if you notice anything.

-Attila

On Fri, Mar 23, 2018 at 8:08 PM, Daniel Pono Takamori <po...@apache.org> wrote:
> Ambari Devs,
> It appears your PR job in Jenkins has had some 20 odd jobs in queue
> due to pinning your job to the 'hadoop' label.  I'm unsure why you've
> tied your job to that label but I've changed it to the more generic
> 'ubuntu'.  The Hadoop nodes are dedicated to main Hadoop projects and
> since we've lost some build capacity we need those nodes for the long
> standing jobs otherwise the queue grows to unmanageable levels.
> We are looking into getting more capacity to alleviate these problems
> and will keep you updated.  It might also be worth looking into a way
> to get your jobs to run in parallel when applicable, since having 20
> PRs need to run serially isn't that convenient.
>
> Cheers,
> -Pono on behalf of Infrastructure