Some links and notes for those knowing nothing about osgeo infrastructure (like me): http://wiki.osgeo.org/wiki/Infrastructure_Transition_Plan_2010 http://wiki.osgeo.org/wiki/AdhocVM http://osuosl.org/ http://www.linux-kvm.org If I got it, osgeo3 and osgeo4 are hosted in OSU OSL using KVM managed by Ganeti with vCPU/CPU 1.75 and 1.5 (i.e 14 and 12 virtual CPU on 8 physical cores).
On Sun, Apr 8, 2012 at 1:11 PM, Tim Sutton <[email protected]> wrote: >>> above now. Possibly we could ask OSGEO for another vm if needed. >> >> Yes, this seems the best option to host Radim benchmarks, right? > > Yes - I almost have all the tests passing for QGIS now. After that I > want to set up automated testing so that as soon as something gets > broken we are all made aware of it. It would be nice to have a place > to do that. I know little about virtualization. I worry that it may be difficult to get information about real physical resources and configuration changes within a VM. The number of VM/vCPU may change while /proc/cpuinfo on a VM remains the same. It seems to me, that for the benchmarks, it would be necessary to pin VM vCPU (and only those vCPU) to reserved physical CPUs: http://docs.ganeti.org/ganeti/master/html/design-cpu-pinning.html . I am not sure if that would be accepted by OSGEO SAC. If we ask for dedicated CPUs, we should be able to document that CPUs will be used most of time. Nightly build + tests + benchmarks running on the same VM could guarantee such effective CPU usage. Radim _______________________________________________ Qgis-developer mailing list [email protected] http://lists.osgeo.org/mailman/listinfo/qgis-developer
