On 9 September 2014 14:45, Levente Uzonyi leves@elte.hu wrote:

Yes, we're running out of disk space on box3. I think the best would be to ask the SFC for more disk[1].

That would be ideal, if possible.

I particularly _don't_ want to have the CI jobs using VMs outside their workspace. Right now, you can look at a build, and find out immediately what VM it's using. You can upgrade the VMs used by CI any time you want, without interrupting other services on the box. And so on.

If we have to conserve disk space, we can do things like delete unused artifacts (especially VMs) after we're finished, at the cost of increased network traffic and build times. The CI scripts can bootstrap their environment (except for external dependencies like ruby and cmake). We just need to be careful then not to delete _useful_ artifacts like updated images, test results, benchmark reports, and so on.

Also jenkins is getting rather old on the box. Is it still necessary to stick to the currently installed version[2]?

It'll always be a balancing act between just having something work, and keeping up with the bleeding edge. I opted to have stuff just work, but perhaps once I've finished this current round of CI hacking, we should upgrade Jenkins.

frank

Levente

[1]https://www.gandi.net/hosting/iaas/disk_resize [2]http://lists.squeakfoundation.org/pipermail/box-admins/2013-July/001414.html

On Tue, 9 Sep 2014, Frank Shearar wrote:

...

Hi,

I thought it high time to break up a mega-job (http://build.squeak.org/job/ExternalPackages/) into (some of) its subparts, because the mega-job takes far too long to run, and makes it hard to understand the current state of play in the Squeak ecosystem.

A side effect of this is more disk usage. Please let me know if I need to get more creative with my disk usage.

Thanks,

frank