Hi,

I have some servers that use 90% or 100% CPU during some high load processor and I often get CPU warning in PRTG.

I wanted to configure for the alarm to be generated if they continued for after one hour with 90% collections.

I tried to adjust the check interval but it keeps generating the alarm after the first 90% identification, even with the setting to only warn after five intervals.

Thanks.

Article Comments

Hello Sandro,

Thank you for your message.

Regarding what you would like to achieve, I'm afraid that the option If a Sensor Query Fails can't help here as it only concerns requests which do not work. For example, if multiple requests fail the sensor will trigger an error in x interval.

In the current case, to avoid receiving alerts every time the CPU load reaches the limits set you have the following options:

  • Increase the latency of the state trigger to cover multiple intervals For example, if the scanning interval is set to 1 minute, using a latency of 300 seconds would trigger an alert only if the sensor reached the limit during that time.

If you have questions, do not hesitate.

Regards.


Nov, 2021 - Permalink

Hi,

that's good, i see an alternative.

However I was left with some doubts if when setting this configuration the sensor would use this condition to change the alarm status keeping it green during this peak period.

1 - Should I continue using (Triggers that can be inherited from parent objects) or should I use (Use only the triggers defined above)?

2 - I liked the option (Limit trigger), but when I set it for example:

When the channel "Total (%)" goes above "100" for at least "3600" seconds, execute (Email and push notification to administrator group.

Will this cause the sensor to remain "green" until it reaches "3600" at 100% cpu?

Thanks.


Nov, 2021 - Permalink

Regarding your first question, this is completely up to you. If the notification trigger you desire to have is already configured at the parent device and inherited by the sensor(s), then you have nothing to do. Otherwise, you can indeed use your own triggers by selecting the option you mentioned and creating it/them afterwards.

Concerning the threshold trigger, it will only send an email based on the channel value (if its reaches the limit for the period specified), no matter if the sensor is in error or not. Of course, you can still have limits configured to have the sensor(s) going in warning/error state and only use a threshold trigger to get the alert.

Regards.


Nov, 2021 - Permalink

Hi,

yes I have a default setting inherited for all devices.

As I now have a specific need for some servers I was in doubt whether adding a threshold trigger would conflict with the one inherited from the parent or would it override the default.

My need is not the email, but the red alarm in PRTG. Because we use PRTG mobile and I want to prevent these known cases from appearing in PRTG in the period and condition I defined as mentioned above.

Thanks.


Nov, 2021 - Permalink

When you add notification triggers to a device which already inherits triggers from its parent, the latter will simply be added to them. Therefore, it means that you will receive an email for each trigger configured (if their conditions are fulfilled).

If hiding the warning/error states is what you are looking for, then I can only invite you to disable/remove the limits in the channels settings. I'm afraid that the triggers do not influence the sensor(s) state here, consequently it's not possible to trigger an error (excepted a simulated one) when the condition are fulfilled (ex: 100% load over 3600 seconds).

Regards.


Nov, 2021 - Permalink