OVA deployment and configuration

[paul.higgsboyo]

Hi,

I have been trying to deploy a 5000 seat CUPS 8.6.3 server on a UCS201M2 (fully licenced) that already has 3 x 7500 seat CUCM servers on it.

The CUPS install fails every time while trying to install the database (cant remember the exact spot).

As a test I re-deployed as a 2000 seat CUPS and it installed fine.

I raised a TAC case and explained that I needed CUPS to be 5000 seat and the TAC engineer has suggested that I just expand the HDD to 250Gig.

I have asked if making this change will make the server unsupportable but so far I havent had a reply.

Anyway, I backed up the server, stopped it and edited the server settings to match those of a 5000 seat OVA CUPS, powered the server back up and its working fine. Has anyone else been in a similar situation before and have you adopted the same "fix"? Here are the OVA hardware reservations:

Offical OVA CUPS 2000 seat server is built as follows:

1 vCPU (1500 MHz reserved)

4 GB RAM with 4 GB reservation

1x 80 GB vHDD

Offical OVA CUPS 5000 seat server is built as follows:

2 vCPU (2500 MHz reserved)

4 GB RAM with 4 GB reservation

2x 80 GB vHDD

My CUPS 5000 seat server is built as follows:

2 vCPU (2500 MHz reserved)

4 GB RAM with 4 GB reservation

1x 250 GB vHDD

Looking forward to hearing opinions/experiences.

Regards,

Paul.

[Jonathan Schulenberg]

I have to admit, I find this "solution" to be more than a little surprising. The most important detail of the OVA is that it includes pre-aligned vDisk files. GNU/Linux does not properly align its sector alignments to the VMFS by default. Without proper sector alignment every I/O operation within the VM will result in two operations on the actual storage. This will half the storage performance and cause iowait for all applications running on that C210 chassis, not just CUP.

So, the question that needs answering is whether the steps you took retained the sector alignment or not. I'm not aware of a way to check this since we do not have root access to the VM although it is possible to have vSphere ESXi rebuild the vDisk images manually to get them aligned.

My advise to you would be to push back on TAC: Something else is going on here. I watched a coworker install the same CUP 8.6(3) on the same C210M2 server using the same ESXi hypervisor yesterday and it worked. If that is truely the workaround than the TAC engineer had better be able to cough up a defect ID to justify it.

[paul.higgsboyo]

Hi Jonathon,

Thanks for your reply.

I'm no expert with VMWare but I guess that the secotor alignment should be OK as the server was deployed initially via an OVA, albeit I have extended the HDD after the server was deployed.

I have gone back to TAC to express these concerns though and to ask for confirmation that all shoulb be OK.

[paul.higgsboyo]

Update to this.

I raised Jonathans' concerns to Cisco TAC who advised that deploying a server via OVA, building it and then amending the disk size is not supported. We can however deploy a server, edit the disk size and then install the OS/application.

This is what I have done.

However, deploying the server via OVA for 2000 users and extending the HDD to 250gig still resulted in a failed installation again bombing out at the database post install stage.

I have rebuilt again successfully on a standard OVA for 2000 users but this still doesnt help as I need to build for 5000 users.

I have passed the case back to TAC and am waiting to hear from them.

Paul.