Hi everyone, our company has two servers running 3.0.2 of ESX server and 2.0.2 virtual centre.
We encountered a problem whereby we were not able to deploy any new virtual machines to these two ESX servers (ESX A & ESX B) that were in a DRS/HA cluster (ESX-AB).
The initial symptom was that on completion of the New virtual machine wizard the Create Virtual Machine task would hang in the Progression state and eventually time out.
As part of trying to diagnose the fault the cluster configuration was first removed. The remove cluster task hung at 33% progress for ESX A and ESX B completed and removed from the cluster, as further fault diagnosis we found that the vpxa process was not running on ESX A, so to restart this process we removed ESX A from Virtual Centre and attempted to add the host back to virtual centre. The cluster config is now removed.
Now when we try to add the host back to virtual centre, Virtual centre complains that we are not licensed to add the host even though we are. I can use VI client to connect to ESX A and I can then unlicense the ESX server which reflects in the FLEX LM license server however we still cannot get ESX A into the virtual centre server inventory. We have tried restarting the FLEX LM license service and Virtual Centre Server services but this does not make any difference. We have tried restarting ESX A but it still complains about the licensing, we've tried removing the ESX hosts and the datacentre they reside on but it still won't come into virutal centre.
After sending all the support logs to VMware, their support seem to think that the issue is related to duplicate virtual machines being registered in the hosts and they've asked me to identify any duplicate VMs on the two hosts that may be resgitered somehow, secondly they've suggested that I unregister all the VMs from both ESX A and ESX B which will cause us major downtime as you cannot unregister a powered on VM.
My thoughts on this are that we should restore the VC database back to a point where the cluster was unconfigured as I beleive the problem is that the current database has VMs registered in a HA cluster that now does not exist and thus the database is currently corrupt. Alternatively we restart with a new database.
My question is, does the the DRS/HA cluster reside solely on the database and does the ESX server not have any clue that it is part of a cluster - my thoughts are that this is right as the clustering is actually a part of Virtual Centre rather than the ESX servers?
When I use VI client to conenct directly to ESX A and create a VM it says the VM is being created on a HA datacentre rather than a the host ESX A - very confusing.
If anyone has any advice on this I would really appreciate it as a second opinion as I'm really tempted to restore the database back to when the cluster was configured.
Regards,
Graeme
