Inconsistent poweron failures saying parent disk missing

I've opened a support ticket on this (SR# 12169721004) but I thought I would ask here too.

Thanks,

Wayne

Power on failures:

I have failures when I power on a successfully deployed configuration.  About half of the VMs fail claiming that the parent disk cannot be found. Configurations are deployed to a single host.

All VMs are in the same datastore and were deployed from a library. The VMs in question have one link in the disk chain because the library was consolidated.

That same library has successfully deployed a configuration and that configuration has successfully deployed and powered on all VMs.

Furthermore, If I undpeloy the failed configuration and immediately  re-deploy it a second time, VMs that failed to power on the first time will successfully power on the second time and machines that successfully powered on the first time might fail the second time.

For example a machine named “mmnfs” successfully powered on in the first attempt and failed to power on in the second attempt.

Here is a sample error

at LabManager.Backend.DRSClusterDeploymentBucket._AsyncDeploy(CDeployableVMContainer container, CVirtualMachine[] vms, IAkimbiManagedServer[] deploymentServers, DeployOptions options, FenceRuntimeMap nfMap, Boolean cfgDeployed, NetIpMap netIpMap, RollbackList rollbackList)

• Failed to power one or more virtual machines.
  

•    at LabManager.Backend.DeploymentBucket.PowerOnVMs(CVirtualMachine[] vms, DeployOptions options)
  

   at LabManager.Backend.DRSClusterDeploymentBucket._AsyncDeploy(CDeployableVMContainer container, CVirtualMachine[] vms, IAkimbiManagedServer[] deploymentServers, DeployOptions options, FenceRuntimeMap nfMap, Boolean cfgDeployed, NetIpMap netIpMap, RollbackList rollbackList)

•    at LabManager.Backend.CAkimbiTask.WaitTillDone(Boolean throwException, Int32 millisecondsTimeout)
  
• 
     at LabManager.Backend.CAkimbiTask.WaitTillDone(Boolean throwException)
  
• 
     at LabManager.Backend.CAkimbiTask.WaitTillDone()
  

   at LabManager.Backend.DeploymentBucket.PowerOnVMs(CVirtualMachine[] vms, DeployOptions options)

Database badly fragmented:

Our Lab Manager performance is poor and the windows machine running the LM server appears fine.  The LM documentation does not talk about DB maintenance and I have not yet found any on the forums (Plese point my at that info if I'm blind).

I do find that when list the deployed VMs there is a 100% CPU spike on the Windows server. I have run the SQL studio management tool to list the physical indexes and they are badly fragmented. I assume this relate to the above power on failures.

I found several procedures for performing maintenance on a 2005 server express that I have applied to an isolated clone of my lab manager server. Those procedures claim to have run successfully but do not defragment or reindex anything.

One of the procedures I used is  http://www.sqldbatips.com/showarticle.asp?ID=27