Hallo,

the main goal of the use case here is to get an alarm, if an application hangs.

In my test case I have got an application that I have monitored for some days now via an WMI process sensor. So I have data within the channels "prozessor usage", "number of handles" and so on.

The application always does something and never sleeps so that over a period of time (measured in 60 sec intervals) I could see that "processor usage" is between 0 and 12 percent and it is not longer than 2 intervals in a chain at 0 percent.

When the application hangs, the processor usage stays at 0 percent for more than 2 intervals.

So I decided to put an lower error limit on the "processor usage" channel of 1 percent and on the sensor level I configured, that after 2 errors the sensor should go in "Alarm" state.

But as you know, this does not work because the PRTG help says, errors of a channel always switch the sensor immediate into error state.

So has anyone an idea how to solve my problem, to get an error after 2 following intervals processor usage an 0 percent?

Regards

Frank

Article Comments

Dear FrankNFurter

While you cannot directly set a notification to X consecutive scans with a certain value, you can add a threshold trigger to get notified when a channel has a particular value for a certain amount of time.


Jul, 2015 - Permalink

Dear Mr. Seifert,

your idea is good, but in our company at the moment we decided not to use notifications. It seems that nobody here wants to check the tickets every hour nor wants to get more emails.

What we want is an error state and an alarm in circumstances where we need it.

Any other idea how to solve the problem?


Jul, 2015 - Permalink

This kind of error status definition is not possible with PRTG. In principle, it would be possible to still use notifications, but not sending an email and running a script instead which changes the status of another custom sensor. This involves manual programming.

It would also be possible to create a factory sensor, set it to alarm via custom status formula and pause it. The notification could perform an API call to resume that sensor, so it turns red.


Jul, 2015 - Permalink