Hyper-V Failover Cluster Validation – think twice
I was talking to some guys with a quite solid VMware background and they were working on setting up a Hyper-V cluster with Windows Azure Pack as frontend.
Their environment was a traditional setup with Hyper-V hosts connecting to a SAN with FC thus using CSV´s for storing the VM´s. All VM´s including their management was configured on these hosts.
As a best practice they ran the Failover Cluster Validation (and yes you should run it or you won’t get support) but as I will show you, they had an unwanted restart of their VM´s including the one they had failover cluster manager started so look and think twice before starting that validation with all tests!
with no HA – VM´s there are not so much to think about but once you add VM´s there is, and it can have some consequences…
If you in the wizard add the CSV volume to the test you can see the warning showing that you will get the VM´s stopped doing the validation,
Going next in the wizard and then going back the warning has disappeared!
But of course in the Confirmation it still says that the VM roles will be stopped but if you are in a hurry and start the validation then all VM´s on the CSV volumes will be stopped and then started and maybe that was not what you had intended from the beginning!
So look and select the options carefully when doing a validation of a cluster with production workloads in 🙂
Comments
Same thing applies when you do this on a storage space cluster – if you opt to validate storage the validation will offline the storage pools during the validation.
I keep a separate 1GB CSV specifically for running cluster validation without interrupting hosted HA-VMs when validating using Failover Cluster Manager. (1GB may actually be far larger than needed for this purpose.) FCM will allow you to select the specific CSVs to use for validation, unlike SCVMM which does not provide the opportunity to choose a specific volume.