Cluster availability report

[A_Mikkelsen]

Hi,

I'm looking for a script that reports a clusters availability/uptime during the last month, been looking but haven't been able to find any.

I prefere the script to integrate into vCheck, but thats not required.

does anyone have or know of such a script ?

br

A. Mikkelsen

[LucD]

Can you be a bit more specific on what kind of availability you want to report on ?

Is it the availability of the ESXi nodes in the cluster only ? That would then just need to measure the uptime of each of the ESXi nodes.

Or the availability of sufficient resources to run all the VMs in the cluster ?

Or ...

[Zsoldier]

That's a somewhat loaded question since a cluster is made up of several hosts and is in itself a 'virtual' object.  In other words, a cluster could technically be considered up by simply just existing.  It doesn't necessarily have to have any hosts associated w/ it.  That's why, at least in 5.0, there is no 'uptime' metric for a cluster.  You would have to figure out how you want to measure that from the individual hosts and possibly provide an average.

So something like this:

$Stats = Get-Cluster NameofClusterYouWantToMeasure | Get-VMHost | Get-Stat -Stat sys.uptime.latest -Start ((get-date).addminutes(-10))

$AvgUptimeinDays = ($Stats | where {$_.value -ne "0"} | Measure-Object -Property Value -Average).Average / 86400

If your average uptime in days was longer than 30, than you were probably 'up' for the last month.

[A_Mikkelsen]

I'm looking for a way to show the availability of a cluster in regards to (Last month):

1. Sufficient cluster resources to run the VM's
2. Hosts downtime in minutes (Scheduled/failure)
3. VM's downtime in minutes (Scheduled/failure) - a reboot with in the OS is to be seen as Scheduled

Ideally I like to show the above graphically

[LucD]

How would you be able to distinguish host and VM downtime as scheduled or failure ?

Is that info available somewhere ?

[A_Mikkelsen]

The info is not directly available, but it is in the event db.

My initial take on how to create this report is:

Downtime caused by a failure will trick an HA event.

For a VM:

• A power on event will occur after the HA event - so downtime will be the time from the HA event to the power on event (alternative until Tools status -eq "running") **

For a host:

• A host connected event will occur after the HA event - so downtime will be from the HA event to the connected event

Scheduled downtime, could be calculated as:

For a VM:

• From the time of a "shutdown" event to the power on event (alternative until Tools status -eq "running") **h

Because I can't come up with an suggestion on how to calculate downtime from a "restart" or "reset" event, I have to leave it out of the report.

For a host:

• From host a hosts successful "Enter maintenance mode" event to host a successful "Exit maintenance mode" event

**) Not sure that the event db has/hold an entry for when a VM's tool status -eq "Running", so this might not be possible.