VMM 2012 – missing VMs do not appear in VMM console

In VMM 2012 situations may arise where a VM is no longer visible in the VMM console, note that I am not referring to a VM that is present but in a ‘missing’ state, that is something different.

VMM relies heavily on the SQL backend. If something is missing it could be caused by a duplicate VMID as it was in my case.
Firstly, find the VMID of the VM that is missing, you can find this by looking at the XML file of the VM.
Then do a search via powershell on the VMM server for the VM:
Get-SCVirtualMachine -ID VMID
If this comes back with error 801 then the VM could exist already in the database.
locating the VM is difficult. Try searching for the hostname of the VM:
get-SCVirtualMachine | Where ComputerName -eq Hostname
Examine any VMs that are returned. If this is an environment where cloning takes place on a regular basis, then it could be that an improperly cloned VM has somehow retained its original VMID, or it could be a library object. Finding and removing the missing/orphaned culprit will fix the issue.

SCVMM 2012 SP1 – Bare Metal Build Error 803d0008 A quota was exceeded

During a VMM bare-metal build error 803d0008 was showing up on the host console. The error occurred when the Host was attempting to send the discovered data back to the VMM server.

error2

The logs showed the following:

error1

 

Right at the bottom the error that stood out was “The numbers of bytes written exceeded the specified quota of 65536 bytes)

It looked as if there was perhaps to much information discovered and VMM was not able to deal with additional data over the 65536 byte limit.

This was a rebuild of an existing host, I had recently performed bare-metal builds successfully the day before.

The host happened to have a large number of LUNs presented to it from the SAN (23 to be exact).

The solution for me was to un-present the LUNs.

 

VMM 2012 SP1 – Error 26874 when applying logical switch to host (or during bare metal build)

Error (26874)
This operation is not permitted since uplink port profile set (Uplink_Port_Profile) in physical adapter on host would go out of scope for host

Recommended Action
Delete the logical switch instance on the affected host(s) and retry the operation

—————————————————————————————————————–

This error will occur when you are trying to assign a logical switch to a host that is in the wrong host group.

The host group must be associated with the logical switch you are trying to apply to the host.

 

 

SCVMM 2012 & r2

I noticed a client was getting these errors when live migrating VMs between cluster nodes:
Error 23801 “No available connection to the selected VM Network can be found”
Error 23811 “Ports are not available for VM Subnet”
These errors would occur despite there being a VM Network available on both nodes. The Live migration would work, but these errors were frustrating.

This can occur when there are additional NICs on the host that have the “Available for Placement” box selected, even if those NICs are disabled. The solution is to untick this box for each host.

Live Migrations Fail between SCVMM 2012 SP1 hosts

While performing some cleanups for a client I noticed a problem where certain live migrations were failing.

I knew I had been performing updates on various hosts to bring them in alignment. Upon closer inspection I could see that the Virtualization Software Version was changing on hosts that had been updated:

Virtualization Software Version Mismatch Clean zoom

 

In the above snip you can see the host in Maintenance mode has a different version of the virtualization software. The version field is talking about the Hyper-V version, rather than the VMM agent version.

As usual, when something fails in SCVMM often the next step is to try it in failover cluster manager, results below:

FailoverClusteringMigrationError

Unfortunately because I had been updating the hosts while they had running VMs on them (and then placing them into maintenance mode once they were ready to reboot) I was left with one host that would not allow live migrations.

The solution?

I had to use Quick Migration to move the VM’s, while this is not technically live, it was quick enough to get the job done.

Lessons learned? Place servers into maintenance mode before you start patching! and keep all your hosts at the same level.

 

Also it is worth pointing out that there have been some particularly damaging patches released by MS recently, rather than explain it myself I will point you to Aidan Finn’s blog where he talks about the problematic patches.

SCVMM 2012 Bare-Metal Failing on HP G7 Blade (IPMI Failure)

Found an issue with a blade yesterday where SCVMM 2012 was failing to power on the blade. The communication was occurring over IPMI. This was the first G7 that I had been testing bare-metal deployments on (having previously worked with Gen 8 Blades).

The iLO event logs would reveal the following entries:

IPMI/RMCP login by scvmm – x.x.x.x(DNS name not found)

IPMI/RMCP logout by scvmm – x.x.x.x(DNS name not found)

 

Whats more, after this occurred the ilo remote console would become unresponsive, requireing the blade to be reseated, or a reset of iLO.

 

Interestingly there was an option on the iLO Dedicated Network Port  that specified that iLO client applications (does this include IPMI?) should use IPv6 first

iLo

 

The fix for me was to disable IPv6 on the iLO dedicated network port, then SCVMM could communicate with the blade via IPMI.