How To Add an Additional Windows Perfmon Counters to Resource Manager – Zenoss

Applies To

Resource Manager 4.2.x
Microsoft Windows Zenpack 2.1.2 and later

Summary

Resource Manager administrators may encounter the need to monitor data points offered by Windows systems that are not monitored by Resource Manager by default. This article explains the method for adding additional data points by (i) confirming that target systems offer the data point and (ii) by adding the data point to a Resource Manager monitoring template.

Procedure

Confirming The Counter Is Available

Log on to a target server and open a command prompt. If necessary, at the prompt, use the cd command to change to a directory where the logged on user has write privileges. For example:
cd c:\users\kbauthor\documents
Type the following at the prompt to generate a list of available PerfMon counters on the system and to output the list to a text file named counters.txt in the directory specified in the step above:
TypePerf.exe –q > counters.txt
Using a file browser, open the text file, then identify the full path of the desired PerfMon counter. In this document the counter \Terminal Services\Total Sessions is used as an example.

Note: The procedure outined in this document calls for modification of one or more Resource Manager monitoring templates. Modification of the default (built in) templates in Resource Manager is not recommended. Instead, overriding templates locally to individual servers or to device subclasses is recommended when template modifications are being made. This document follows this best practice by detailing the override of the default Windows template to an individual server. For more information on overriding templates, see Templates in the the Zenoss Service Dynamics Resource Management Administration document.
In the Resource Manager web interface, navigate to the target server and click on the server name.
On the lower left of the screen, click the Action Wheel (the gear shaped icon) and choose Override Template Here.
From the drop down list that displays, select Device (/Server/Microsoft).
Click Submit.
Click the small arrow next to Monitoring Templates on the left side of the screen to expand the list of templates bound to the device.
Click the newly overridden template named Device (Locally Defined).
Click the + (plus) symbol under Data Sources to add a new data source.
Populate the Name field and choose Windows Perfmon from the dropdown list for Type.
Submit the form.
Double click on the new data source and populate the Component field with a name or phrase that is relevant to the data point (in this example, Terminal Services). The value entered here will appear in any events generated by Resource Manager for the data point.
Populate the Counter field with the full name of the Perfmon counter. In this example, the value is \Terminal Services\Total Sessions.
Submit the form.
Click on the name of the new data source to highlight it, then click the Action Wheel (gear shaped icon) below Data Sources.
Choose Add Data Point.
Populate the Name field.
IMPORTANT: THE NAME CHOSEN FOR THE DATA POINT MUST BE THE SAME NAME AS THE DATA SOURCE. For example, if TerminalServicesTotalSessions is the name of the data source, the data point must be named TerminalServicesTotalSessions.
Submit the form.

Creating a Graph (optional)

If desired, complete the following steps to create a graph to visually depict the performance of the data point over time:

On the right side of the screen click the + (plus) symbol underneath Graph Definitions.
Populate the Name field in the form that appears and submit the form.
Click the name of the graph to highlight it and click the Action Wheel (gear shaped icon) underneath Graph Definitions.
Choose Manage Graph Points.
Click the + (plus) symbol to add a graph point.
Choose Data Point from the list that displays.
From the dropdown menu, choose the new data point created in the previous steps.
Submit the form.
Click Save.
Double click the name of the graph to open the View and Edit Graph Definition form.
Populate the Units, Min Y and Max Y fields.
Appropriate values for this example would be:
- Sessions for the Units value
  (so administrators know they are looking at the number of Terminal Services sessions)
- 0 for the Min Y vaue
  (so the lowest value possible in the graph will be 0 sessions )
- 10 for Max Y
  (sets the highest values depicted on the graph)
Verify that the data point is successfully being polled by waiting a sufficient amount of time for the several monitoring poll cycles to be completed (often 20 minutes or more if the default polling cycles have been retained). If a graph is created, click on Graphs on the left side of the screen to view graphs for the device. The graph should visually depict a value, and the values for cur: av: and max: at the bottom should be something other than -nan.

Troubleshooting

If the graph fails to show values after three to four times the normal polling cycle time, troubleshoot your data point by completing the following:

Log on to the command line of the Resource manager host and change to the zenoss user:
su - zenoss
Run a single poll cycle against the target server to verify that the new data point is being polled:
zenpython run -v 10 -d [device name]

NOTE: in the above command, device name refers to the device's name in Resource Manager, which may or may not correspond to its actual host name.
Search the output that results for confirmation that the new data point is being polled. In the example detailed in this document, a line such as the following would indicate that the data point is being polled:
2014-10-16 10:29:15,200 DEBUG zen.RRDUtil: /opt/zenoss/perf/Devices/10.177.215.251/TerminalServicesTotalSessions_TerminalServicesTotalSessions.rrd: 2.0, @ 1413469755

If the output does not include details of the data point being polled, double check the settings above. Pay particular attention to verification that the Data Source and Data Point names are identical.