Check Windows Performance Monitor Counters

                                        
Plugin for Icinga/Nagios that allow to check a group of Windows performance counters

                    

        
        

            
            

                

    

    
Check Win Perfmon


Plugin for Nagios/Icinga that allows to check a group of Windows performance counters specified in a XML file.


Checks value of performance counters based on thresholds specified in XML.


Returns exit code and performance data in Nagios/Icinga format.


Performance output


Download Check Win Perfmon v2.0. 


Release 2.0 has breaking changes with 1.4. You must change 'auto' parameter in xml files for 'automemory', 'autodisk' or 'autonetwork'


Please read below before use it! 


Preconfigured XML files


In downloaded zip package, there are several .xml files preconfigured:


    

  • 

    PerfMonNetwork.xml: Performance Counters to check network load.
    

    • 

  • 

    PerfMonPhysicalDisk.xml:Performance Counters to check physical Disk load.
    

    • 

  • 

    PerfMonCPU.xml: Performance Counters to check CPU load.
    

    • 

  • 

    PerfMonMem.xml: Performance Counters to check Memory (RAM and virtual) load.
    

    • 

  • 

    PerfMonMSQL.xml: Performance Counters to check Microsoft SQL Server.
    

    • 

  • 

    PerfMonWebService.xml: Performance Counters to check Microsoft IIS Web Service.
    

    • 

  • 

    PerfMonPrinter.xml: Performance Counters to check Microsoft Print Server.
    

    • 

  • 

    PerfMonCB.xml: Performance Counters to check Microsoft Connection Broker Server and its WID.
    

    • 

  • 

    PerfMonHyperV.xml: Performance Counters to check Microsoft Hyper-V Server.
    

    • 

  • 

    PerfMonWID.xml: Performance Counters to check Microsoft Windows Internal Database of WSUS.
    

    • 



Usage


check_win_perfmon.exe [parameters]:


    

  • 

    -f, --xmlFile        (Default: perfcounts.xml) XML file with performance counters configuration.
    

    • 

  • 

    -s, --maxSamples     (Default: 3) Amount of samples to take from perfmon.
    

    • 

  • 

    -t, --timeSamples    (Default: 1000) Time between samples in ms.
    

    • 

  • 

    -n, --noalerts       (Default: false) Always returns 0. Useful to get only performance data without alerts.
    

    • 

  • 

    -p, --xmlParams      (Default: none) Array of params to change in xml file. Read below for examples.
    

    • 

  • 

    -v, --verbose        Verbose output for debuging.
    

    • 



Check performance counters of PerfMonMem.xml taking 10 samples with 2 sec interval.


check_win_perfmon.exe -f PerfMonMem.xml -s 10 -t 2000


Examples


Example CPU counters: check_win_perfmon.exe -f PerfMonCPU.xml


OK - All performance counters between range | 'ProcessorTime'=3%;95;100;0;100 'UserTime'=2%;85;95;0;100 'DPCTime'=0%;15;20;0;100 'InterruptTime'=0%;10;15;0;100 'ProcessorQueueLength'=0;4;8;;


Example Memory counters: check_win_perfmon.exe -f PerfMonMem.xml


OK - All performance counters between range | 'CommittedBytesInUse'=57%;80;90;0;100 'AvailableMBytes'=4083MB;1024;512;0;8192 'AvailableMBytesPercent'=50%;13;6;0;100 'FreeSystemPageTableEntries'=2867405056;5000;4000;; 'PagesSec'=0;5000;6000;;


Example Physical Disk counters: check_win_perfmon.exe -f PerfMonPhysicalDisk.xml


OK - All performance counters between range | 'AvgDiskSecTransfer'=0.0002s;0.04;0.05;; 'CurrentDiskQueueLength'=0;32;40;; 'AvgDiskSecWrite'=0.0002s;0.04;0.05;0; 'AvgDiskSecRead'=0s;0.04;0.05;0; 'IdleTime'=100%;20;15;0;100


Example Network counters: check_win_perfmon.exe -f PerfMonNetwork.xml


OK - All performance counters between range | 'BytesTotalSec'=1885.7051B;15728640;17825790;0;20971520 'BytesTotalSecPercent'=0%;75;85;0;100 'OutputQueueLength'=0;2;3;;


Example Microsoft SQL counters: check_win_perfmon.exe -f PerfMonMSQL.xml


OK - All performance counters between range | 'TotalServerMemory'=8381528KB;14680060;16252930;0;16777220 'TotalServerMemoryPercent'=50%;88;97;0;100 'TargetServerMemory'=8388608KB;14680060;16252930;0;16777220 'TargetServerMemoryPercent'=50%;88;97;0;100 'PageReadsSec'=0;90;100;; 'PageWritesSec'=0;90;100;; 'BufferCacheHitRatio'=100;95;90;0;100 'BufferCacheHitRatioPercent'=100%;95;90;0;100 'PageLifeExpectancy'=109982.6641;400;300;; 'LazyWritesSec'=0;15;20;; 'FreeListStallsSec'=0;1;2;; 'MemoryGrantsPending'=0;1;2;; 'BatchRequestsSec'=16.6571;1000;2000;; 'UserConnections'=115.3333;600;700;; 'LockWaitsSec'=0;1;2;; 'ProcessesBlocked'=0;1;2;;


Creating new XML files to check your own performance counters


You can set up your own performance counters adding them to xml files or creating new ones.


To list available performance counters on a system in a PowerShell console type:


# Get all counters
Get-Counter -ListSet * | Select-Object -ExpandProperty Counter
# Get specified counter 
Get-Counter -ListSet *processor* | Select-Object -ExpandProperty Counter


You can check performance counters on a Windows system: Start Menu->Administrative Tools->Performance Monitor->Clic on plus symbol


XML Format


XML file used must have the following format, for example:


<?xml version="1.0" encoding="UTF-8" ?>

        Processor
        % Processor Time
        _Total
        ProcessorTime
        %
        95
        100
        0
        100

        Memory
        Available MBytes
        none
        AvailableMBytes
        MB
        1024
        512
        0
        auto

        Hyper-V Virtual Machine Health Summary
        Health Critical
        none
        VirtualMachineHealthCritical
        none
        none
        gt1
        none
        none



Warning: Counter names must be in english.


In the example above, program will check two counters. For each counter, we need to set:


    

  • category: Category of performance counter
    • 

  • name: Name of the performance counter.
    • 

  • instance: Instance of performance counter. Some performance counter does not have instance, in this case the value must be: none. 

      

    • autonetwork: detects main network interface.
      • 

    • autodisk: detect system disk.
      • 

    • 

  • friendlyname: name of performance counter which program returns in performance output.
    • 

  • units: units program returns in performance output. Check units on Nagios performance data docs.
    • 

  • warning: Warning threshold for performance counter.
    • 

  • critical: Critical threshold for performance counter.
    • 

  • min: minimum value of performance counter. If you do not know the minimum value, it has to be: none.
    • 

  • max: maximum value of performance counter.  If you do not know the maximum value, it has to be: none.

      

    • autonetwork: detects network interface speed in kb/s.
      • 

    • automemory: detects system memory.
      • 

    • 



If max and min are specified, program returns one more performance result for calculated percent value.
Max and min must have different value.


Warning and critical can be a % of max value. For example:


    80%
    95%
    0
    20480


If you want to check only warning or critical threshold, it should have the format: lt or gt, and none for not checked one.
For example, warning if counter is less or equal than 15:


    lt15
    none


Critical if counter is greater or equal than 90% of max:


    none
    gt90%
    0
    20480


XML with parameters


To avoid creating an xml file for each network interface, disk or sql instance, we can create a generic xml file with parameters, for example.


<?xml version="1.0" encoding="UTF-8" ?>

        Network Adapter
        Output Queue Length
        {0}
        OutputQueueLength

        {1}
        {2}
        none
        none

        Network Adapter
        Output Queue Length
        {3}
        OutputQueueLength

        {2}
        {5}
        none
        none



To pass parameters:


check_win_perfmon.exe -f PerfMonNetworkParams.xml -p "Interface 1" "1" "2" "Interface 2" "5"


Params of type {n} will be replaced in order.


XML file would be:  


<?xml version="1.0" encoding="UTF-8" ?>

        Network Adapter
        Output Queue Length
        Instance 1
        OutputQueueLength

        1
        2
        none
        none

        Network Adapter
        Output Queue Length
        Instance 2
        OutputQueueLength

        2
        5
        none
        none



You can use two parámeters in same field too.


<?xml version="1.0" encoding="UTF-8" ?>

        PhysicalDisk
        Current Disk Queue Length
        {0} {1}:
        CurrentDiskQueueLength

        32
        40
        none
        none



To check it:


check_win_perfmon.exe -f PerfMonDiskParams.xml -p "1" "D"


check_win_perfmon.exe -f PerfMonDiskParams.xml -p "3" "F"


System Load


I tried to minimize system load during program execution, but check performance counters allways has an impact on system performance. Program execution has a 5% of CPU usage on old systems and a minimun impact on modern servers. The more performance counters you check at a time, the more system impact.


Icinga Agent Configuration


Command


object CheckCommand "check_win_perfmon" {
    import "plugin-check-command"
    command = [ "C:\\Program Files\\ICINGA2\\sbin\\check_win_perfmon.exe" ]
    arguments = {
        "-f" = {
            value = "$xml$"
            order = 1
            description = "XML file"
        }
        "-t" = {
            value = "$interval$"
            order = 2
            description = "Time between samples"
        }
        "-s" = {
            value = "$samples$"
            order = 3
            description = "Samples to take"
        }
        "-n" = {
            value = "$noalerts$"
            order = 4
            description = "Always return 0"
        }
        "-p" = {
            value = "$xmlParams$"
            order = 5
            description = "XML params"
        }
    }
}


Service


apply Service "CPU Load" {
    import "generic-service"
    check_command = "check_win_perfmon"
    vars.xml = "C:\\Program Files\\ICINGA2\\sbin\\PerfMonCPU.xml"
    command_endpoint = host.name
    assign where host.vars.os == "Windows"
}

apply Service "Network Load" {
    import "generic-service"
    check_command = "check_win_perfmon"
    vars.xml = "C:\\Program Files\\ICINGA2\\sbin\\PerfMonNetwork.xml"
    command_endpoint = host.name
    assign where host.vars.os == "Windows"
}

apply Service "Disk_0 Load" {
    import "generic-service"
    check_command = "check_win_perfmon"
    vars.xml = "C:\\Program Files\\ICINGA2\\sbin\\PerfMonPhysicalDisk.xml"
    command_endpoint = host.name
    assign where host.vars.os == "Windows"
}

apply Service "Memory Load" {
    import "generic-service"
    check_command = "check_win_perfmon"
    vars.xml = "C:\\Program Files\\ICINGA2\\sbin\\PerfMonMem.xml"
    command_endpoint = host.name
    assign where host.vars.os == "Windows"
}

apply Service "SQL Server Load" {
    import "generic-service"
    check_command = "check_win_perfmon"
    if ("msql2012" in host.vars.checks) {
        vars.xml = "C:\\ProgramData\\icinga2\\var\\lib\\icinga2\\api\\zones\\global-templates\\_etc\\scripts\\PerfMonMSQL2012.xml"
    }
    if ( "msql2014" in host.vars.checks) {
        vars.xml = "C:\\ProgramData\\icinga2\\var\\lib\\icinga2\\api\\zones\\global-templates\\_etc\\scripts\\PerfMonMSQL2014.xml"
    }
    if ("msqlnamed1" in host.vars.checks) {
        vars.xml = "C:\\ProgramData\\icinga2\\var\\lib\\icinga2\\api\\zones\\global-templates\\_etc\\scripts\\PerfMonMSQL2014Params.xml"
        vars.xmlParams = "NAMED1"
    }
    if ("msqlsharepoint" in host.vars.checks) {
        vars.xml = "C:\\ProgramData\\icinga2\\var\\lib\\icinga2\\api\\zones\\global-templates\\_etc\\scripts\\PerfMonMSQL2012Params.xml"
        vars.xmlParams = "SHAREPOINT"
    }
    if ("msqlnamed2" in host.vars.checks) {
        vars.xml = "C:\\ProgramData\\icinga2\\var\\lib\\icinga2\\api\\zones\\global-templates\\_etc\\scripts\\PerfMonMSQL2014Params.xml"
        vars.xmlParams = "NAMED2"
    }
    vars.samples = "10"
    vars.noalerts = ""
    command_endpoint = host.name
    assign where (regex("^msql?[a-z0-9]+",host.vars.checks,MatchAny))
}


References


Values and counters are based on System Center Operations Manager checkins. You can check it out here.


Values and counters for Microsoft SQL are based on articles from SLQ Shack and Database Journal.


Updated tresholds based on the amazing tool PAL created by Clint Huffman from Microsoft.