LucD notes

When you are running multi-node vSphere clusters, you probably already had the experience that one or more of your LUNs were not visible on all nodes. Now you can try to find out which LUN is missing on which node the hard way through the vSphere client. Or you can use the force of PowerCLI and run a script that will report all this in a handy spreadsheet.

To make the script as flexible as possible it should be able to handle any n-node cluster. And as you some of you might know, the Export-CSV cmdlet has some problems with variable length rows. Luckily there is a handy solution I already used in my yadr – A vDisk reporter post.

param($clusName,$csvName=("C:\Temp\" + $clusName + "-LUN.csv"))

$rndNum = Get-Random -Maximum 99999

$LunInfoDef = @"
	public string ClusterName;
	public string CanonicalName;
	public string UsedBy;
	public string SizeMB;
"@
$LunInfoDef = "public struct LunInfo" + $rndNum + "{`n" + $LunInfoDef

$esxServers = Get-Cluster $clusName | Get-VMHost | Sort-Object -Property Name
$esxServers | %{
	$LunInfoDef += ("`n`tpublic string " + ($_.Name.Split(".")[0]) + ";")
}
$LunInfoDef += "`n}"

Add-Type -Language CsharpVersion3 -TypeDefinition $LunInfoDef

$scsiTab = @{}
$esxServers | %{
	$esxImpl = $_

# Get SCSI LUNs
	$esxImpl | Get-ScsiLun | where {$_.LunType -eq "Disk"} | %{

		$key = $esxImpl.Name.Split(".")[0] + "-" + $_.CanonicalName.Split(".")[1]
		if(!$scsiTab.ContainsKey($key)){

			$scsiTab[$key] = $_.CanonicalName,"",$_.CapacityMB
		}
	}

# Get the VMFS datastores
	$esxImpl | Get-Datastore | where {$_.Type -eq "VMFS"} | Get-View | %{
		$dsName = $_.Name
		$_.Info.Vmfs.Extent | %{
			$key = $esxImpl.Name.Split(".")[0] + "-" + $_.DiskName.Split(".")[1]
			$scsiTab[$key] = $scsiTab[$key][0], $dsName, $scsiTab[$key][2]
		}
	}
}

# Get the RDM disks
Get-Cluster $clusName | Get-VM | Get-View | %{
	$vm = $_
	$vm.Config.Hardware.Device | where {$_.gettype().Name -eq "VirtualDisk"} | %{
		if("physicalMode","virtualmode" -contains $_.Backing.CompatibilityMode){
			$disk = $_.Backing.LunUuid.Substring(10,32)
			$key = (Get-View $vm.Runtime.Host).Name.Split(".")[0] + "-" + $disk
			$scsiTab[$key][1] = $vm.Name + "/" + $_.DeviceInfo.Label
		}
	}
}

$scsiTab.GetEnumerator() | Group-Object -Property {$_.Key.Split("-")[1]} | %{
	$lun = New-Object ("LunInfo" + $rndNum)
	$lun.ClusterName = $clusName
	$_.Group | %{
		$esxName = $_.Key.Split("-")[0]
		$lun.$esxName = "ok"
		if(!$lun.CanonicalName){$lun.CanonicalName = $_.Value[0]}
		if(!$lun.UsedBy){$lun.UsedBy = $_.Value[1]}
		if(!$lun.SizeMB){$lun.SizeMB = $_.Value[2]}

	}
	$lun
} | Export-Csv $csvName -NoTypeInformation -UseCulture
Invoke-Item $csvName

Line 1: The definition of the parameters that can be used with the script. Note that the second parameter takes a default when is not present.

Line 3: If you would run the script multiple times from the same PowerCLI prompt you would get an error saying “Add-Type : Cannot add type. The type name ‘LunInfo’ already exists.“. To avoid this error the script generates on each run a random number between 1 and 99999, which it appends to the type name. This should, unless you have thousands of clusters avoid the error message.

Line 5-19: Since the number of nodes is unknown before the script is actually running, I create a new type, called LunInfon (where n is the random number generated before).

Line 14-16: The new type will contain a property for each node in the cluster. The name of this property is the first qualifier of the FQDN of the host.

Line 21: The script keeps all the information it will gather in a hash table called $scsiTab. The key into the hash table is composed from short hostname and the Canonical name of the LUN.

Line 26-33: From the information provided by the Get-ScsiLun cmdlet, the script will populate the hash table. In this loop the CanonicalName and the Capacity of the LUN are stored in the hash table.

Line 36-43: From the Datastore object the script finds out for all datastores in the cluster which LUN they use.

Line 45-55: The script runs through all the virtual machines on the cluster and checks all virtual disks if they are RDM backed. If the script encounters a RDM it stores the name of the virtual machine and the label of the hard disk in the hash table.

Line 57-68: All the information in the hash table is now grouped per Canonical name. For each Canonical name the script creates a LunInfon object and stores the information from the hash table in that object.

Line 69-70: The array with LunInfon objects is exported to a CSV file and the CSV file is opened.

You save the script as a .ps1 file. In the example screenshots below I used the name Cluster-LUN-list.ps1.

Start the PowerCLI prompt and make sure that you are connected to the vCenter.

From the PowerCLI prompt you can call the script with the name of a cluster as a parameter.

In this case the script will create the CSV file as C:\Temp\MyCluster17-LUN.csv.

Or you pass the name of the CSV file as a second parameter to the script.

In this case the script will store the report in LunRpt.csv on the root of the D-partition.

Note that some local disks seem to have the same Canonical name on all nodes in a cluster. I experienced that with some IBM RAID Controllers. In that case the entries for those disks will not be correct in the report. In itself this is not really a problem since those disks are local and no LUNs from the SAN.

The following screenshot shows a report made for a 4-node cluster.

You can clearly see which LUNs are:

• only visible on 1 of the nodes
• used as RDM disks
• used for datastores
• still available

This report will surely make it easier to talk with your Storage Administrator