Every once in a while I ‘m asked by a customer to troubleshoot issues on clusters (windows clusters or Hyper-V clusters… no difference for the purpose of this post). One of the first things I do is to getting information of which generating and gathering the cluster log files.

I used to logon to every server, create the file, copy it to my laptop and when I had all the files I could start troubleshooting… as you may suspect, this can take some time and is basically ‘manual labor’. How can I automate this a little?

With the following command (run on a node in the cluster) you generate the log files for each node… but are saved on the node you are logged on to. This saves you connecting to all the nodes one by one and generating the files by hand, etc.

cluster /cluster:clustername log /gen /copy:.

But… this is done with the use of cluster.exe which is to disappear in a future Windows Server release. How can we do this in PowerShell?
Well… rather easy actually

Get-Cluster | foreach {get-clusterlog –destination D:\Temp}

Last weekend I was at a customer that was very enthusiastic about virtualizing their servers with Hyper-V.
They immediately started to build their clusters and virtualized their servers. They were rather shocked when cluster resources were not brought online when a node failed, migrations did not work, etc. Also, SCVMM shown a status of ‘Unsupported cluster configuration’ for the VM’s that were experiencing issues.

The first thing I do in situations like these is to run a cluster validation report, which shown me that either something was changed after the cluster was created or no cluster validation was run at all.

Next, I look for similarities… which can become a journey in itself  
The Hyper-V Manager told me that nothing was wrong with the virtual machine configurations… and Failover Cluster Manager also reported no errors.
Although this customer had SCVMM installed and configured, I was not allowed access to the console… but that’s a discussion for another day.
I still wanted to see if the issues were related to a specific host or cluster… or if there were VM’s without this status.
Since I had no access to the SCVMM console, I tried the SCVMM PowerShell Module… and guess what? I did have access to it (since I can do way more with PowerShell compared to the management console I’m gonna have a nice discussion with the IT guy that denied me access to the console…).

Here’s the command to get the VM names, status en node they are running on:

Get-VM | select name, vmhost, status | sort status | convertto-html | out-file D:\Temp\VMStatusList.html

And the output was a follows:

As you can see, my guess was right… the VM’s running on the HV-cluster were just fine but those running on the HYPER-cluster were reporting the ‘Unsupported Cluster Configuration’ status.
Just to make sure, I migrated a VM to the HV-cluster… and the status changed from ‘Unsupported Cluster Configuration’ to ‘Running’.

So, I logged on to the HYPER1 and HYPER2 node and walked through the basics for a cluster configuration.
Suddenly everything became clear… 4 NICs in one team for all the network traffic (yes, this includes the heartbeat traffic), VHD’s stored on both CSV and on local disk for the same virtual machine, etc.
So we’ve rebuild the cluster and everything is running smoothly

Now what have we learned by all of this?
Step 1: Make sure you have a clear problem definition.
Step 2: Get access to the tools and servers you need.
Step 3: Check if the problem still exists
Step 4: Run a cluster validation report and solve any warnings/errors.
Step 5: Look for similarities
Step 6: ……………..

Every once in a while I find myself in an environment where no one has a clue which VM’s are part of a network, let’s say the DMZ.
Since this is mostly identified by IP address/range I create a script to provide me with a list of IP configurations by using the SCVMM cmdlets:

Get-VMMServer localhost | out-null
$VMs = Get-VM
Foreach-object ($VM in $VMs)
{
   if (($VM.ComputerName -ne $NULL) -AND ($VM.ComputerName -ne "*"))
   {   
      $IPS = [System.Net.DNS]::GetHostAddresses($VM.Name)
      trap [System.Exception] {"The following virtual machine is not registered in DNS:"; continue;}
      write-Host ""      
      write-Host "VM:   " $VM.Name
      write-Host "Name: " $VM.ComputerName
      Write-Host "IP:   " $IPS
      Write-Host ""   
   }
}

As you may have noticed, this script uses DNS to resolve the IP address.

Now, with the new PowerShell cmdlets introduced with SCVMM 2012 life suddenly becomes much, much easier… especially for me since I´m a huge fan of one liners
With SCVMM 2012, use the following to get the same information:

Get-SCVirtualMachine –VMMServer localhost | Format-List -property Name, ComputerName, IPV4Address

Every once in a while I go to a customer that has no clue what operating systems they are running inside their virtual machines. But when this is a datacenter managed with SCVMM, one simple line of PowerShell code is all I need to get the information I desire:

Get-VM –VMMServer SCVMM | Select-Object Name, OperatingSystem | ConvertTo-HTML | Out-File D:\Temp\VMOSList.html

This will provide a nice list with VM names and their operating systems:


Note that this command will look at the VM as they are known within SCVMM, not inside the VM!

When migrating, deploying or converting P2V in SCVMM, by default those file transfers are done by BITS over HTTPS:

This is the cause of some overhead and when your company policy does not require you to use encrypted file transfers for VM deployment, you could just turn it off… right?

When you select the properties of your Hyper-V hosts (by right-clicking ‘All Hosts’) you can select ‘Allow unencrypted file transfers’.

This will enable you to use the HTTP instead of the HTTPS profocol for the file transfer. So with a P2V or (quick/live) migration you will see that BITS over HTTP instead of BITS over HTTPS is used.

But… you may notice that the deployment through a template is still using BITS over HTTPS.
Why is this method still using HTTPS instead of HTTP? The other methods picked the change up perfectly, so why doesn’t this one?

Well… there is a little gotcha.
For the SCVMM Library, where your template is stored, you can set the ‘Allow unencrypted file transfers’ independently from the setting on the hosts!
So, when you goto the properties of the SCVMM Library (by right-clicking your library server) you can configure the setting

After configuring this, do a new deployment and you will notice that BITS over HTTP is used which causes a very nice decrease of the time it takes to deploy a VM

One if the category of ‘useful PowerShell one-liners’… getting the maximum size of dynamic VHD’s:

Get-VirtualHardDisk -VMMServer SCVMM | Select-Object @{Label=”Name”;Expression={$_.Name}},@{Label=”VHD Maximum Size”;Expression={[Math]::Truncate(($_.MaximumSize-$_.Size) / 1GB)}} | Where-Object {($_.VHDType -Match “DynamicallyExpanding” -And $_.Name -NotMatch “{“)} | Export-Csv DynamicVHDSizes.csv

This will provide an output like the following:

I found this very handy when calculating if the available storage has been over subscribed

But let’s go a little bit further… some time ago I wrote a post about getting the available amount of space within dynamic VHD files. What if we were to combine that script with the one above?
Then we would get the amount of increase in storage usage when converting your dynamic VHD’s into static VHD’s
Here we go:

Get-VirtualHardDisk -VMMServer SCVMM | Select-Object @{Label=”Name”;Expression={$_.Name}},@{Label=”Current Size”;Expression={[Math]::Truncate($_.Size / 1GB)}},@{Label=”VHD Maximum Size”;Expression={[Math]::Truncate(($_.MaximumSize-$_.Size) / 1GB)}} | Where-Object {($_.VHDType -Match “DynamicallyExpanding” -And $_.Name -NotMatch “{“)} | Export-Csv PrepareVHDConversion.csv

And this will provide an output like:

In my previous post I’ve shown howto use a PowerShell one-liner to create a simple inventory of the memory configuration and availability of your Hyper-V hosts and virtual machines.

Now, let’s go a little more global. When creating a design for a (new) Hyper-V environment that information is not something you would use… one would need some high level statistics instead of a bunch of raw data.
Again, this is where PowerShell comes in:

Get-VM -vmmserver localhost | measure-object -property memory -sum -average -maximum –minimum

With the measure-object cmdlet you can easily get some statistics… the above one-liner provided me the following output:

Now let’s explain this a little since this is, again, just a bunch of numbers…
The count is the number of virtual machines it found.
The average is the average amount of memory configured on the virtual machines.
The maximum is the highest amount of memory configured on a virtual machine.
The minimum is the lowest amount of memory configured on a virtual machine.
The property explains explain itself