[Systems] RFC: Wanted: Servers for build farm

David Farning dfarning at sugarlabs.org
Sun Apr 5 12:06:53 EDT 2009

Looks good.


On Sun, Apr 5, 2009 at 9:40 AM, Sascha Silbe
<sascha-ml-ui-sugar-systems at silbe.org> wrote:
> This is the mail I'm going to send out to sugar-devel. Any comments?
> Is the Prgmr bit
> a) correct and
> b) diplomatic enough?
> Hi all!
> As you might know, Sugarlabs is running a network of build machines for
> automatic and manual building of Sugar (on several distros) and SoaS (on
> Fedora). There are plans to integrate this with further automatic tests
> (e.g. SugarBot [1]) and even ideas like integration with aslo for providing
> precompiled activities (for those with binary blobs) for a variety of
> platforms. So the build network is going to be a vital piece of
> infrastructure.
> The current slaves are hosted by Prgmr Xen Hosting (thanks!). Unfortunately,
> the host software running the VMs is getting too old to be useful (svn
> stopped working on the Fedora Rawhide slave) and we seem to be unable to
> reach Prgmr.
> Since we'd like to extend our farm anyway (so we can test on more distros
> and more hardware), we're now looking for new server sponsors.
> Ideally, we'd have one or more dedicated servers (i.e. with full control
> over the host) for each architecture, in order to decrease maintainance
> overhead, be able to add new targets fast and ensuring all slaves can work
> properly (i.e. host software not getting too old).
> In practice, each single VM/buildslave is welcome (as long as we've got
> sufficient access on the host to control the VM), especially for non-PC
> architectures.
> What the servers need:
> - enough CPU cache (no Celeron/Duron)
> - fast CPU preferred (but not a requirement)
> - enough RAM for all VMs (0.5..1G per VM)
> - enough HD space for all VMs (~10G per VM)
> - stable internet access
> What they don't need (but would be great of course):
> - loads of network bandwidth (an occasional upload of SoaS is expected to be
> the largest transfer)
> - low network latency
> - RAID (all data is already stored elsewhere)
> - loads of HD space (a standard 0.5TB drive should be sufficient for quite a
> while)
> While we're looking for hosted servers (i.e. someone else maintains the
> hardware and provides network access), we'll probably be able to find a nice
> place if you'd like to donate just the hardware.
> [1] http://dev.sugarlabs.org/ticket/594
> [2] http://www.prgmr.com/
> CU Sascha
> --
> http://sascha.silbe.org/
> http://www.infra-silbe.de/
> Version: GnuPG v1.4.9 (GNU/Linux)
> N0Y5d8TTF8DMkqKfNVMg5ss+CxVrlXdpfuJEJKrcqLZ+quU6tRHyFeb4LZ2oWCd1
> kv0N63VJsen7TeOoa6LzkLtRNJDaioQZHtncmhfPizmzpYwSTewDwYo76LgbwmWc
> VIejmOqdgp+/m2sluMamupYjqOPatD7YmE0mZ9Bsq00DmyDm7iZ2Bq4RtfrHuhu3
> KoSk7Ed63uuyhBv00wNv/6ze45fFUWk3yooJC/8xTW+3fpyNGgXECUwYE7SWP58M
> nIf2yxWlOr9VMzp552fGGHUw8QS3ZJpgq2K03gfmcdwDux30hYurf5R1bKDB9bM=
> =MpGO
> _______________________________________________
> Systems mailing list
> Systems at lists.sugarlabs.org
> http://lists.sugarlabs.org/listinfo/systems

More information about the Systems mailing list