vMotion of certain machines would not work

Hi there,

I experienced a frustrating issue today.  I tried to vMotion a virtual machine and it would not go.  Compatibility issue.  Error message is below.

marshall sunburst

I could vMotion with some of the hosts, but not most of them.  This was an EVC problem.  Even in the error message it said that and sent me to a KB article.  But it did not help me.  What EVC mode should I choose?  This is my story of how I dealt with this.

Background

A while back I had a Web Client issue.  It did not work after my 6.0 upgrade.  VMware asked for extended time to access it to try and figure it out.  But a month or two ago they said no choice, but install vCenter over again.  So I did.  But, I forgot to set a EVC mode in my cluster.  Everything worked.  Once it was running, and according to my vSphere Best Practices, I did a vMotion of a VM to all hosts.  However, by happenstance I picked a VM that would in fact vMotion anywhere.  It seems that only the virtual appliances I have could not vMotion most places – that includes Log Insight, CloudPhysics, and vCenter.

Solution

First I had to figure out the mode I would use.  So I looked at each host Summary page and got the processor info.  Things like E5520 or E5-2407.  Then I visited this page. I then located the processor in the CPU Series column of the first table.  I had processors in L2, L3, and L4.  That means I need to use L2 as it is the lowest.  It had a baseline (from the Baseline column) of Nehalem.  So now I know that is the EVC mode I need to use.

Now, I need to put that into implementation and that is the hard part.  There is likely a number of ways to do this.  What I did was this – and it is partially determined by a small lab.

  • Turned off all virtual machines except for vCenter (VCSA) and one domain controller.
  • I tried to enable EVC using the Nehalem mode and I was told that one server could not and I found the DC VM on that server.  So I knew that VM had to be managed.
  • I moved one host to a new cluster and turned it on.
  • I moved that one VM – the DC to that host.
  • I enabled EVC on my main cluster.
  • Using the Embedded Host Client on that recently moved host, I shutdown the DC.  Then I unregistered it.
  • Using the Embedded Host Client on a host that was in the cluster that EVC had been enabled on, I registered the DC and powered it on.
  • I moved the host from the new cluster back to the main one.
  • I powered everything on in the proper order.
  • I now tested my vMotion with something that previously could not successfully vMotion previously and it worked fine!  So all good.

I hope that no one has this problem.  One of us is enough.  But if you do, this process above will work for you.  There is likely other ways to do this as well.  Let me know if you have questions, comments, or suggestions.

And thanks very much to Florian Grehl who did that Intel CPU EVC Matrix as it really helped!

Michael

=== END ===

 

Tagged with: , ,
Posted in Home Lab, How To

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: