For every Windows system there are Performance Counters available which will deliver information and details about the state of the general health of certain components of Windows itself but also disks, drivers, software and network interfaces for example.
To get to know more about them in general, you can have a look on the Microsoft Documentation for Performance Counters.
In order to fetch Performance Counters data by using the Icinga PowerShell Framework, we have to first clarify on how to access counters. Every counter is split into two mandatory components:
- Category
- Counter
Each Category can provide a various amount of different Counter objects you can access to read data out. An example counter is
'\Memory\available mbytes'The notation is always a leading \ followed by the Category, another \ and the Counter itself.
In this example, Memory is the Category and available mbytes is the counter we can fetch data from. It contains available Megabytes for allocation to a process or for system use.
Now in addition to the previous mentioned structure, there is one exception when a counter contains multiple Instances. An Instance is used for example by processor, disk or network interfaces, on which multiple values are stored for different disks, processor cores or network interfaces.
To access a certain Instance you will have to extend the Category with brackets () and write the name of the Instance into it. You can use * to fetch data for all Instances.
'\Processor(*)\% processor time'Please keep in mind that not all counters will require instances.
The way the Icinga PowerShell Framework is designed allows the global usage of the counters with their english name. This means it doesn't matter which language your Windows system is running it. How ever, localised systems will often print localised names of the counters, making it hard to use them on systems running a different language.
To fetch a list of available counters on a specific system with their english names, there is a Cmdlet avaialble in the Framework itself:
Show-IcingaPerformanceCounterCategories;System
Memory
Browser
Cache
Process
Thread
PhysicalDisk
LogicalDisk
...
As we now know which Categories are available on our system, we can have a look on them which Counter are present. For this there is also a Cmdlet avaialble directly within the Framework:
Show-IcingaPerformanceCounters 'Memory';\Memory\committed bytes
\Memory\pool nonpaged bytes
\Memory\page writes/sec
\Memory\transition faults/sec
...
Note that the Cmdlet will output the counters with the correct notation already which means you can simply copy & paste them into your code for usage. The same applies for counters with instances:
Show-IcingaPerformanceCounters 'Processor';\Processor(*)\dpcs queued/sec
\Processor(*)\% c1 time
\Processor(*)\% idle time
\Processor(*)\c3 transitions/sec
...
Last but not least you can fetch all available Instances for a Performance Counter with a custom Cmdlet introduced with Icinga PowerShell Framework version 1.2.0:
Show-IcingaPerformanceCounterInstances '\Processor(*)\% processor time';Name Value
---- -----
_Total \Processor(_Total)\% processor time
0 \Processor(0)\% processor time
1 \Processor(1)\% processor time
Not only will it print you the name of the Instance but also the full path to use for fetching the value of a specific Performance Counter.
By using New-IcingaPerformanceCounterArray, you can provide multiple Performance Counters and fetch them all at once. This is the recommended way, as some counters will require a certain Sleep interval to have valid data available. By using the array handling, all counters will be loaded and initialised at once, reducing the overall required sleep duration to an absolute minimum.
You can add multiple Performance Counter paths to the Cmdlet. All of them will be fetched at once;
New-IcingaPerformanceCounterArray '\Processor(*)\% processor time', '\Memory\committed bytes', '\Memory\available mbytes';Name Value
---- -----
\Processor(*)\% processor time {\Processor(_Total)\% processor time, \Processor(0)\% processor time, \Processor(10)\...
\Memory\available mbytes {error, sample, type, value...}
\Memory\committed bytes {error, sample, type, value...}
You can access the values directly by using the specific Performance Counter as index:
$Counter = New-IcingaPerformanceCounterArray '\Processor(*)\% processor time', '\Memory\committed bytes', '\Memory\available mbytes';
$Counter['\Processor(*)\% processor time'];Name Value
---- -----
\Processor(_Total)\% proces... {error, sample, type, value...}
\Processor(0)\% processor time {error, sample, type, value...}
\Processor(10)\% processor ... {error, sample, type, value...}
\Processor(20)\% processor ... {error, sample, type, value...}
...
Now you can access the values with their Instance path:
$Counter['\Processor(*)\% processor time']['\Processor(_Total)\% processor time'];Name Value
---- -----
error
sample System.Diagnostics.CounterSample
type Timer100NsInverse
value 0
help % Processor Time is the percentage of elapsed time that the processor spends to execu...
And of course the value itself
$Counter['\Processor(*)\% processor time']['\Processor(_Total)\% processor time'].value;Of course you can also access a single Instance of a Performance Counter or using a Performance Counter not having any Instances at all:
$Counter = New-IcingaPerformanceCounterArray '\Processor(_Total)\% processor time';
$Counter['\Processor(_Total)\% processor time'].value;Plenty of Counters provide Instances, which are required to allow the specific assignment of values to an object. Such examples are Network Interface Counters for different interfaces or PhysicalDisk/LogicalDisk counters for each installed disk.
By simply fetching all informations from a counter, like \LogicalDisk(*)\free megabytes, we receive each disk value within our output. This is fine in general, is however problematic in the following scenario:
$Counter = New-IcingaPerformanceCounterArray '\LogicalDisk(*)\free megabytes', '\LogicalDisk(*)\% idle time';Of course we will receive the correct amount of data, but the output is not something we can effectively work with, because the parent of our access is always the specified counter with all instances below. You will have to access both of these objects to receive informations of your C disk for example:
$Counter['\LogicalDisk(*)\free megabytes']['\LogicalDisk(C:)\free megabytes'].value;
$Counter['\LogicalDisk(*)\% idle time']['\LogicalDisk(C:)\% idle time'].value;As you can see, the problem is not that big, but imagine you have 10 counters you wish to include for your disk. This is not something you wish to handle manually.
As we just learned in general this usage is fine, requires how ever some effort to display and use Counter properly. To resolve this, we will create a structured object now with automatic sorting of the data.
To make the example a little more interesting, lets do as much automation as possible.
At first we will create a variable and store all Counters inside the system provides:
$CounterList = Show-IcingaPerformanceCounters 'LogicalDisk';The Show-IcingaPerformanceCounters Cmdlet is returning all available Performance Counters. We can now use this output and create an array with our New-IcingaPerformanceCounterArray Cmdlet:
$Counters = New-IcingaPerformanceCounterArray $CounterList;Now we have initialised all Counters and Instances and loaded them into our variable. This is quite fine, but we still have the same result as described within our problem scenario:
Name Value
---- -----
\LogicalDisk(*)\free megabytes {\LogicalDisk(HarddiskVolume30)\free megabytes, \LogicalDisk(R:)\free megabytes, \Log...
\LogicalDisk(*)\% disk read... {\LogicalDisk(HarddiskVolume1)\% disk read time, \LogicalDisk(_Total)\% disk read tim...
\LogicalDisk(*)\avg. disk w... {\LogicalDisk(HarddiskVolume13)\avg. disk write queue length, \LogicalDisk(R:)\avg. d...
\LogicalDisk(*)\disk transf... {\LogicalDisk(_Total)\disk transfers/sec, \LogicalDisk(G:)\disk transfers/sec, \Logic...
...
Now let the Framework do the magic and organice the entire Counter hashtable based on a group parent. As we fetch LogicalDisk data, lets group them by the Instances of the category and add our previous loaded counters to the call
$Result = New-IcingaPerformanceCounterStructure -CounterCategory 'LogicalDisk' -PerformanceCounterHash $Counters;Inside our $Result variable we have not sorted all Performance Counters properly based on their Instance and every single Counter is now available by the name of the Counter itself:
Name Value
---- -----
Name Value
---- -----
HarddiskVolume13 {avg. disk queue length, % free space, avg. disk sec/transfer, avg. disk bytes/read...}
HarddiskVolume15 {avg. disk queue length, % free space, avg. disk sec/transfer, avg. disk bytes/read...}
HarddiskVolume30 {avg. disk queue length, % free space, avg. disk sec/transfer, avg. disk bytes/read...}
_Total {avg. disk queue length, % free space, avg. disk sec/transfer, avg. disk bytes/read...}
HarddiskVolume1 {avg. disk queue length, % free space, avg. disk sec/transfer, avg. disk bytes/read...}
G: {avg. disk queue length, % free space, avg. disk sec/transfer, avg. disk bytes/read...}
R: {avg. disk queue length, % free space, avg. disk sec/transfer, avg. disk bytes/read...}
HarddiskVolume24 {avg. disk queue length, % free space, avg. disk sec/transfer, avg. disk bytes/read...}
C: {avg. disk queue length, % free space, avg. disk sec/transfer, avg. disk bytes/read...}Our $Result variable is a simple hashtable which contains all above mentioned data. By using the Instance as first index and the Counter as second index, you can directly access the values:
$Result['C:']['avg. disk queue length'].value;
$Result['C:']['% disk time'].value;
$Result['C:']['avg. disk bytes/transfer'].value;