If you have installed the latest release of PowerCLI (6.3 R1) then you may have noticed a warning if you have run the Get-ESXCLI cmdlet that looks a little like this:
What does it mean?
Whilst talking to the community and listening to our customers we heard that there were a couple of issues with this cmdlet, some that lead to problems with scripts breaking when being used against environments that have more than one version of ESX or maybe upgraded versions of ESX using older scripts with this cmdlet. Lets take a bi of a deeper dive into these issues so you can understand why we made the change and how the new –V2 version is so much better!
Lets discuss the 2 main problems….
1. Positional values and the $null variable
When using the Get-ESXCLI cmdlet we would only allow invocation based on positional parameters, this was somewhat painful as providing just 1 parameter for a esxcli which had 15 options would result in long, weird looking scripts with a lot of $null parameters to fill the blanks, it was hard to work out which parameters where needed and where they fit in between the $null entires. Lets take a look at the “ESXCLI network diag ping” namespace for example, to call this namespace and ping a given address twice in ESXCLI you would need to call the namespace with a –count and a –host option as per the docs here. We would have had to find out where to put these when calling the Get-ESXCLI cmdlet and where to add the $null for everything we didn’t want to use.
Once done we would end up with a script which may look like the following:
$esxcli = Get-ESXCLI -VMHost (Get-VMhost | Select -first 1)
$esxcli.network.diag.ping(2,$null,$null,“10.0.0.8”,$null,$null,$null,$null,$null,$null,$null,$null)
Not the most intuitive way to do things but it worked right?! Well, yeah kind of but that leads us to problem 2…
2. Mixed environments or upgraded environments
One of the rules for updating ESXCLI namespaces is that you are allowed to add options but not remove or modify existing ones, this stops breaking changes for anyone using ESXCLI in a script, the issue was that when people were using these through PowerCLI and potentially using it against multiple hosts of different versions we could have a different number of $null parameters as we need one for every parameter we don’t use.
For example, we may have the following example in ESX 5.1:
$esxcli = Get-ESXCLI -VMHost (Get-VMhost | Select -first 1)
$esxcli.network.diag.ping(2,$null,$null,“10.0.0.8”,$null,$null,$null,$null,$null,$null,$null,$null)
This may work fine but then what if we at VMware extending this namespace to give it an extra option, we would need an extra $null through Get-ESXCLI, so for example in ESX 5.5 this may be:
$esxcli = Get-ESXCLI -VMHost (Get-VMhost | Select -first 1)
$esxcli.network.diag.ping(2,$null,$null,“10.0.0.8”,$null,$null,$null,$null,$null,$null,$null,$null,$null)
As you can see, if we ran either the 5.1 version against a 5.5 hosts or the 5.5 version against the 5.1 host we would have errors and broken scripts. Of course you could check for each version but this starts to get complex when sharing scripts for others to use and the multiple versions of esxcli namespaces we have out there.
So what does V2 do that’s different?
The Get-ESXCLI cmdlet has now been updated with a –V2 parameter which supports specifying method arguments by name, essentially we are able to fix both issues above and provide an easier way to explore what an ESXCLI namespace requires and allow us to provide only the items we care about, lets walk though the same example but using the –V2 switch
Firstly lets store the new version in a variable like we did above:
$esxcli2 = Get-ESXCLI -VMHost (Get-VMhost | Select -first 1) -V2
Next we can find out what options are available to set for the namespace by calling the new CreateArgs method and storing this in a variable:
$arguments = $esxcli2.network.diag.ping.CreateArgs()
Now we can see what’s in the variable by calling it:
PowerCLI C:\> $arguments
Name Value
—- —–
host Unset, ([string], optional)
wait Unset, ([string], optional)
df Unset, ([boolean], optional)
interval Unset, ([string], optional)
ttl Unset, ([long], optional)
debug Unset, ([boolean], optional)
nexthop Unset, ([string], optional)
count Unset, ([long], optional)
netstack Unset, ([string], optional)
size Unset, ([long], optional)
ipv4 Unset, ([boolean], optional)
ipv6 Unset, ([boolean], optional)
interface Unset, ([string], optional)
As you can see, its helped us out here, it has prefilled this array with the options we can specify, now all we need to do is assign values to the entries we want to use:
$arguments.count = 2
$arguments.host = “10.0.0.8”
If we look at what’s in the variable again we can see we have set just the two entries we need:
PowerCLI C:\> $arguments
Name Value
—- —–
host 10.0.0.8
wait Unset, ([string], optional)
df Unset, ([boolean], optional)
interval Unset, ([string], optional)
ttl Unset, ([long], optional)
debug Unset, ([boolean], optional)
nexthop Unset, ([string], optional)
count 2
netstack Unset, ([string], optional)
size Unset, ([long], optional)
ipv4 Unset, ([boolean], optional)
ipv6 Unset, ([boolean], optional)
interface Unset, ([string], optional)
Next all we need to do is send these options back by using the Invoke method and providing our set $arguments variable:
$esxcli2.network.diag.ping.Invoke($arguments)
lets put this together so you can see how easy this is now:
$esxcli2 = Get-ESXCLI -VMHost (Get-VMhost | Select -first 1) -V2
$arguments = $esxcli2.network.diag.ping.CreateArgs()
$arguments.count = 2
$arguments.host = “10.0.0.8”
$esxcli2.network.diag.ping.Invoke($arguments)
Conclusion
As you can see, not only is this much easier to understand but we also don’t have to use lots of $null entries and it will also work against multiple versions of ESXi or allow you to share your scripts knowing that they will work against any version of ESXi that supports the namespaces you used in ESXCLI.
A Win, Win situation here, so why wouldn’t you use the –V2 option?!