Hilton Lange over at the offical SCVMM Blog has posted a fantastic write-up of the mechanics of Dynamic Optimization.
Of particular interest to me is the logic used to determine whether or not to rebalance a cluster. I’ve seen clusters before where Network Optimizations have been enabled and DO does not balance any VMs at all however live migrations work. This is caused by warnings generated during the evaluation process as a result of Network Optimization not being supported on the hardware and as you can see from the above link, any warnings means a host is discounted. Unfortunately there isn’t much information in SCVMM itself about the results of the logic when DO runs; vSphere is a bit better in that regard.
While on an installation of a pair of HP 5900 switches in an IRF stack, I was somewhat taken aback by the size and amount of packaging for the 40GbE QSFP-QSFP Direct Attach Cables [DACs] we were using. Judge for yourself:
During a recent cluster migration from Hyper-V 2008 R2 SP1 to Hyper-V 2012 (more on that another time) I ran into an interesting issue.
Everything was going well except some machines would not start in Failover Cluster Manager and oddly the machine was not visible in Hyper-V Manager. The following events were logged in the event log:
The message from event 21502:
‘VirtualMachineName Configuration’ failed to register the virtual machine with the virtual machine management service.
Cluster resource ‘VirtualMachineName Configuration’ of type ‘Virtual Machine Configuration’ in clustered role ‘VirtualMachineName Resources’ failed.
The Cluster service failed to bring clustered service or application ‘VirtualMachineName Resources’ completely online or offline. One or more resources may be in a failed state. This may impact the availability of the clustered service or application.
Clustered role ‘VirtualMachineName Resources’ has exceeded its failover threshold. It has exhausted the configured number of failover attempts within the failover period of time allotted to it and will be left in a failed state. No additional attempts will be made to bring the role online or fail it over to another node in the cluster. Please check the events associated with the failure. After the issues causing the failure are resolved the role can be brought online manually or the cluster may attempt to bring it online again after the restart delay period.
Event 21502 obviously explains why the VM isn’t showing in Hyper-V Manager and also why it wasn’t possible to edit the settings of the machine:
In Failover Cluster Manager, right-click the VM and click Remove. This will return the machine in Hyper-V Manager however it still won’t start – you’ll get the following error:
Editing the settings of the VM will show that the configuration is invalid:
Fix the network settings and the machine will start.