Public bug reported: We are running eucalyptus 1.6.2-0ubuntu27 on lucid beta1. I will retest this bug on the latest-and-greatest as soon as that is feasible on our cluster.
We have been running many tests with the ROUNDROBIN scheduling policy. As a test I changed it to GREEDY a little over a week ago. The scheduling policy seems to work as expected except that occasionally when servicing large requests the cluster controller will request that a node run more VMs than available cores on a machine. We are using kvm. When making a large request (say 100 VMs) it seems that invariably one of the nodes used to run the machines will be over-subscribed. Our machines have 8 cores each and I often see 9 VMs and occasionally I've seen as many as 12. I have checked that the machines with extra VMs did not have hyper- threading enabled and were therefore reporting the correct number of cores to Eucalyptus according to the /var/log/eucalyptus/euca_test_nc.log file on each system. ** Affects: eucalyptus (Ubuntu) Importance: Undecided Status: New -- GREEDY scheduling policy occasionally over-subscribes nodes (more VMs than cores) https://bugs.launchpad.net/bugs/566715 You received this bug notification because you are a member of Ubuntu Server Team, which is subscribed to eucalyptus in ubuntu. -- Ubuntu-server-bugs mailing list Ubuntu-server-bugs@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-server-bugs