Choosing the correct OVA for deployment… it matters.

I’ve seen a couple customers run into an ugly issue that isn’t pleasant to resolve.

After upgrading their VOS (voice operating system)-based application (CUCM, CUC, CCX, etc.) to a new version the server boots and throws this warning/error:

VMware Installation: 2 vCPU Intel(R) Xeon(R) CPU E5640 @ 2.67GHz, disk 1: 80Gbytes, disk
2: 80Gbytes, 6144Mbytes RAM, ERROR-UNSUPPORTED: Partitions unaligned

The cause is all the cases that I’ve reviewed was the correct OVA not being used in the original deployment:

  1. Deploying without using an OVA at all (VMWare admin building the machine by hand)
  2. Deploying the incorrect OVA for the version of ESX you are on (e.g. vmv7 version with ESXi 5 rather than the correct vmv8 version).
  3. Deploying the incorrect OVA for the version of the application (e.g. deploying the CCX 10 OVA for the build of CCX9 and then upgrading to CCX10).

My process during a migration:

Know which version of the VOS app you are going to install; know the version of ESX you are running on and use the appropriate OVA.

For example if I had a production CUCM system on v9.1(2) and ESXi 5.5 and was replicating the CUCM system to a test environment to take it to v10.5, I must deploy the OVA for CUCM 9.1 vmv8 on the test environment.  I’d DRS out/in and then upgrade the test environment to v10.5.    (I would NOT deploy the v10.5 OVA for the 9.1 migration installation.)  This migration example would not require the use of the 10.5 OVA at all.

Now for CUCM I’d actaully use Prime Collaboration Deployment to make this process MUCH easier (which WOULD use the v10.5 OVA).  But for other apps like CUC or UCCX which don’t have PCD support yet, I’d use that process.

If you get a system with the error

ERROR-UNSUPPORTED: Partitions unaligned

The fix is ugly.

Workaround:
To correct the unaligned disk partitions you must:
1) Take a DRS backup of the system
2) Deploy a new virtual machine from an updated version of the OVA  and
perform a fresh install of the previously-unaligned virtual machine. Details can be found
on this Virtualization DocWiki page:
http://docwiki.cisco.com/wiki/Implementing_Virtualization_Deployments
3) Perform a restore on the freshly-installed node, using the DRS backup taken in step 1

More Info:
If your virtual machine has unaligned disk partitions Cisco cannot provide support for
performance issues. The steps in the Workaround must be followed to put the virtual
machine in a supported state.

Once everything is on v10.x, RHEL6 (used in VOS 10.x)  automatically aligns partitions.  So this should not be an issue in the future once cured on your 9.x system.

4 thoughts on “Choosing the correct OVA for deployment… it matters.

  1. I was wondering if the upgrade to 10 fixes this on its own, or if this ugly workaround process is a pre-requisite for the 10.x migration. I was surprised to find that we inherited of few of these “gems” from a former consultant, and so now have to look at doing this for a CUC and CUCM publisher, both with fairly large databases.

    • Paul, it’s something that you’re going to want to cure before you go to 10. I’m sure you can go to 10, but you’re going to still have that error message pop up, and from what I’ve seen with one customer going to 10 who had the error before the upgraded, it doesn’t cure itself. :-/

      • Is it true that jump upgrades aren’t supported past 9.1.2? I was planning on upgrading virtualized version to 10, DRS out then restore on fresh copy with proper OVA but all documentation I can find references using PCD..

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