Hi everyone,

i'm looking for a way to monitor Services wich are not shown in the "SNMP Windows-Service Monitor" Dialog. WMI is not an Option because of the huge amount of Sensors on one Probe (about 18.000).

Thanks for your Help

Mischa

Article Comments

Hello there,
thank you for your KB-Post.

The following is a Powershell Script that allows you to monitor several services with a single sensor:

So you only need one sensor per server.

Best Regards,
Luciano Lingnau [Paessler Support]


Feb, 2018 - Permalink

Thank you very much. Not exactly what i was looking for, but it will help me.


Feb, 2018 - Permalink

Hi again, i tried your script, but it doesn't work as it should be. The Problem is, even with th script there are about 350 Sensors which is to much for PRTG (more than 100 open requests).

Is there another solution i could try?

Thanks in advance.


Mar, 2018 - Permalink

Hello Mischa,
thank you for your reply.

350 powershell-based sensors may need to be split over multiple remote probes. You also need to carefully plan the scanning interval and make sure these sensors are paused when the host stops responding (Ping as master dependency can be used for this).

Best Regards,
Luciano Lingnau [Paessler Support]


Mar, 2018 - Permalink

Hello Luciano, thanks for your fast answer. These 350 sensors are in 350 different locations connected through WAN. Would it be ok to just add a remote probe to the main location? Another Idea is to add a VM on the main server which only hosts a remote probe. (unique network card is available)

it doesn't seem to me, that the performance of the main system is the Problem (CPU 20% max, over 30 GB of RAM free)

Is there a possibility to just add the Service to the snmp list?


Mar, 2018 - Permalink

Hello Mischa,
thank you for your reply.

QUOTE=These 350 sensors are in 350 different locations connected through WAN.

This can cause issues. I'm not sure how robust Powershell will work over the WAN connections (or what/which ports you will need to allow for this to work (if firewalls are involved). However, as a general rule you're recommended to have remote probes as close as possible to the monitored systems, to reduce the latency in the communication with the monitored device.

Would it be ok to just add a remote probe to the main location? Another Idea is to add a VM on the main server which only hosts a remote probe. (unique network card is available) it doesn't seem to me, that the performance of the main system is the Problem (CPU 20% max, over 30 GB of RAM free)

While PRTG's Core Server is a 64-bit application, the Probe component is a 32-bit process and won't take advantage of more than 4GB of Ram. The number of CPU threads it can make use of can also be misleading on systems with a large number of cores, since it can't use all of them.

Smaller probes (as close as possible to the monitored systems, for example one per remote site) usually work best. However, we also suggest <30 probes per PRTG deployment, so with 350 sites this becomes an issue. In this case you might want to work for probes for a "region".

Is there a possibility to just add the Service to the SNMP list?

Actually, I'm not wondering why you're not seeing the service in the list (not sure why I haven't asked about it earlier). Because basically, the list of services presented by the sensor should list ALL running services from the host (stopped services are not published/visible in the list).

May I ask you for an example of service that doesn't show up in the list when adding the sensor? Does the service's name include special characters? Or is it extremely long maybe?

Best Regards,
Luciano Lingnau [Paessler Support]


Mar, 2018 - Permalink