- for the most part, Windows VM builds time out and fail back to hardware - we can't install VS2010 on the Windows VMs (they're too small) - our ESX hosts are *way* oversubscribed, so running fewer VMs will make the remaining systems (linux/linux64) faster - TinTri hasn't been able to find any silver bullets - most likely, scl3 will not have a VMware cluster How's this for a plan: * Tomorrow we reshuffle the ix slaves (move 12? to try) (bug 673013) * Wait to see how reshuffled, repaired ix slaves handle the load * ??? * profit * reap profits next week, see how new setup is handling load, then start shutting off VMs.
Lots of dupes - we must *really* want to do this.