Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [cross-project-issues-dev] tycho-gmp.gmf.tooling builds consumes enormous amount of memory

On 02/23/2012 05:06 AM, Mickael Istria wrote:
Hi Denis, Winston, all,

Right.  So I propose that the tycho-gmp.gmf.tooling job be moved over to our Sandbox so that its memory consumption does not impact our Hudson instance.  Any objections?
How much trouble does this cause to other jobs? Is this job the only responsible for Hudson being slow? Would removing from Hudson save the world?

The issues we are having with Hudson cannot be isolated to a single job.  It is the sum of many small things that are causing issues.

I am targetting your job to be moved because it is part of that equation -- it stands out as consuming many more times the amount of RAM.  We are also working on many other aspects to improve our Hudson instance -- such as pruning old jobs, pruning old builds (bug 371094), hosting Husdon on Jetty instead of Winstone (367774), restructuring slave1's resources (367772) and upgrading Hudson to a more recent version (371039). 

Just to be clear: none of those issues is the single source of our instability.  But if we do nothing, everyone will continue to be impacted.


 otherwise the CI job for GMF Tooling will be forever in a sandbox, it'd have a bad effect on the project.
I think forever is a strong word, but I don't understand why having a job on the sandbox is bad for the project.  The EGit and JGit used the sandbox for years and managed to blossom their way from incubation into the release train.  Actually, if we cannot get the Gerrit plugin to behave with the production instance, their builds will likely return to the sandbox  :)


Also, be sure that if GMF Tooling job is moved to sandbox, the same problem will happen on sandbox...
Agreed.  However, having the issue manifest itself on the sandbox will not affect 100 other projects -- just yours.



If no, then it seems that the job does not cause real trouble, then let's keep it as it and keep on enjoying life

I've seen enough Hudson-related bugs, complaints, tweets and emails that it is clear to me that no one is enjoying life (wrt hudson.eclipse.org), and that is why we must act.

Thanks for understanding.  Matt will work with your project in order to transition the job over to the sandbox instance with minimal disruptions.

Denis

Back to the top