Stats over time - again

[raymove]

Hi all,

I am trying to write (well, modify) a script that gives the maximum and average CPU and memory stats per cluster over the last week.

I serached quite a bit, and all paths seem to lead to the same thread from 2016 (VMware PowerCLI Forum - VMware {code}) where LucD  seems to have solved the problem with a script that looks like this (I cut the minimum paramters since I don't need them):

$allClusters = @()

Get-Cluster | %{

    $clusstat = "" | Select ClusterName, MemMax, MemAvg, MemMin, CPUMax, CPUAvg, CPUMin

    $clusstat.ClusterName = $_.Name

    $statcpu = Get-Stat -Entity $_ -Start (get-date).AddDays(-30) -Finish (Get-Date)-MaxSamples 100 -Stat cpu.usage.average

    $statmem = Get-Stat -Entity $_ -Start (get-date).AddDays(-30) -Finish (Get-Date)-MaxSamples 100 -Stat mem.usage.average

    $cpu = $statcpu | Measure-Object -Property value -Average -Maximum

    $mem = $statmem | Measure-Object -Property value -Average -Maximum

    $clusstat.CPUMax = $cpu.Maximum

    $clusstat.CPUAvg = $cpu.Average

    $clusstat.MemMax = $mem.Maximum

    $clusstat.MemAvg = $mem.Average

   

    $allClusters += $hoststat

}

$allClusters |

Select ClusterName, MemMax, MemAvg, CPUMax, CPUAvg

Export-Csv $outputfile -noTypeInformation

This script, however, gives nothing in return. When I replace $outputfile with a filew path, the file is empty.

Tried out-gridview, but nothing there either.

This is probably a minor problem, but I cannot for the life of me figure out why I get an empty return here....?

[LucD]

The first thing that comes to mind, is your vCenter keeping the performance data that long.

Have a look at my PowerCLI & vSphere statistics - Part 1 - The basics  post to understand statistical levels, performance data collection and aggregation.

[raymove]

That could be, of course, but changing AddDays(-30) to (-7) yields the same result. I am sure the vCenter keeps data for at least that long.

And I can see these stats in the vCetner advanced performance GUI, just trying to automate the reporting.

[raymove]

I found this little gem from  2008, that works better. It looks similar, but uses a foreach loop:

$allclust = @()

$clust = Get-Cluster

foreach($clu in $clust){

  $clusstat = "" | Select ClusterName, MemMax, MemAvg, CPUMax, CPUAvg

  $clusstat.ClusterName = $clu.name

 

  $statcpu = Get-Stat -Entity ($clu) `

           -start (get-date).AddDays(-7) `

           -Finish (Get-Date) `

            -MaxSamples 10000 `

             -stat cpu.usagemhz.average

  $statmem = Get-Stat -Entity ($clu) `

           -start (get-date).AddDays(-7) `

           -Finish (Get-Date) `

            -MaxSamples 10000 `

             -stat mem.usage.average

  $cpu = $statcpu | Measure-Object -Property value -Average -Maximum

  $mem = $statmem | Measure-Object -Property value -Average -Maximum

 

  $clusstat.CPUMax = $cpu.Maximum

  $clusstat.CPUAvg = $cpu.Average

  $clusstat.MemMax = $mem.Maximum

  $clusstat.MemAvg = $mem.Average

 

  $allclust += $clusstat

}

$allclust | `

  select ClusterName, MemMax, MemAvg, CPUMax, CPUAvg

Gives the following output:

ClusterName : Prod_B

MemMax      : 45,3600006103516

MemAvg      : 45,2667361276592

CPUMax      : 131887

CPUAvg      : 69396,2994011976

ClusterName : Prod_A

MemMax      : 47,0299987792969

MemAvg      : 46,8672157904345

CPUMax      : 108343

CPUAvg      : 61335,8293413174

ClusterName : Management

MemMax      : 41,0499992370605

MemAvg      : 40,9491614952773

CPUMax      : 103128

CPUAvg      : 77024,6107784431

Is there a quick way to get CPU stats in per cent?

[LucD]

Use the metric cpu.usage.average instead of cpu.usagemhz.average

[raymove]

Thank you, you rock.

But then again, you already knew that :smileyhappy: