memory stat cluster

[pierrouille]

Hi,

I'm triyng create a script which gives output as below :

Timestamp                   Total           Usage

05/12/2016 09:00:00     1874756     49,67

05/12/2016 07:00:00     1874760     49,69

12/05/2016 13:00:00     1874760     50,06

05/12/2016 03:00:00     1874759     50,27

05/12/2016 01:00:00     1874759     50,32

total = memory available in GB for this cluster at this time

usage = memory in use in % at this time

Actually result is :

Name               Group

05/12/2016 09:00:00     {1874756, 49,67}

05/12/2016 07:00:00     {1874760, 49,69}

05/12/2016 05:00:00     {1874760, 50,06}

05/12/2016 03:00:00    {1874759, 50,27}

05/12/2016 01:00:00    {1874759, 50,32}

I use following cmd :

Get-Stat -Entity (Get-Cluster Cluster_LINUX_G7) -Stat mem.usage.average,mem.totalmb.average -IntervalSecs 7200 | Group-object Timestamp | select Name, Group

I would like to get ouput in a a csv file (in order to create graph in excel after)

Can you have some ideas ?

Thanks in advance

[LucD · Accepted solution]

Try something like this

$report = Get-Stat -Entity (Get-Cluster Cluster_LINUX_G7) -Stat mem.usage.average,mem.totalmb.average -IntervalSecs 7200 |

Group-object Timestamp | %{

  New-Object PSObject -Property @{

    Time = $_.Name

    Total = $_.Group | where{$_.MetricId -eq 'mem.totalmb.average'} | Select -ExpandProperty Value

    Used = $_.Group | where{$_.MetricId -eq 'mem.usage.average'} | Select -ExpandProperty Value

  }

}

$report | Export-Csv report.csv -NoTypeInformation -UseCulture

[MikkoJ]

There is the proper way to do it (LucD answer), and the ugly:

I did this by oneliner, of which I'm not proud.

For some reasons I did not get return for mem.totalmb.average, hence the workaround by counting total memory of cluster.

get-stat -entity (get-cluster Cluster_LINUX_G7) -stat mem.usage.average -intervalsecs 7200 | select Timestamp, @{n="Total";e={[math]::round((((100-$_.Value)/100)*(((get-cluster gra-standalonehosts |get-vmhost).memorytotalgb | measure-object -sum).sum)),0)}}, @{n="Usage";e={($_.Value)}}

This would run way faster like this:

$clustertotalmemory = ((get-cluster Cluster_LINUX_G7 |get-vmhost).memorytotalgb | measure-object -sum).sum

get-stat -entity (get-cluster Cluster_LINUX_G7) -stat mem.usage.average -intervalsecs 7200 | select Timestamp, @{n="Total";e={[math]::round((((100-$_.Value)/100)*$clustertotalmemory),0)}}, @{n="Usage";e={($_.Value)}}