Processor alerts

Hello Everyone and HNY....for 2023

Hopefully a simple Question & Answer

I'm trying to monitor "CPU Processor alerts", its a simple thing right,

But I wan't to know how to configure the alert to only tell me if a processor is at 95% for a certain time, say 30 minutes..etc.

Thanks

Regards

Find more posts tagged with

processor

cpu_alerts

Accepted answers

All comments

ecmel

You can select tick box for "Condition must exist" in trigger condition and specify the duration to wait before triggering the alert as shown below.

This will check all polling data in last 30 minutes and trigger alert if the condition is true for all polling cycles of a node.

SteveK

We took a similar approach to ecmel with ours. Instead of hardcoding to 90% we use an or statement with critical value reached is true or Current Value (CPU Threshold) is greater than Critical value. We had an issue with a previous version where the critical value reached didn't always update, so we added in the comparison as a backup.

This way you can utilize a single alert and adjust percentages on the node side.

There is also a warning value set as well which we use for a different alert. I want to say the default settings for CPU thresholds are 90% for warning and 95% for critical.

DazM

Hi everyone,

And a big Thank you for the comments and solutions...

I'm clear now (Thanks to your replies) on how to create a specific time trigger alert on a device for CPU.

However if I create this trigger, won't the global CPU alerts carry on working? or does it intelligently remove the device from the global list? (Hope that makes sense..)

The global list I am referring to is the "Manage All Alerts" in All settings.

Thanks

Daz

SteveK

So if you already have an existing CPU alert, they would both trigger. If you have the existing out of box alert on, this would probably be more of a replacement for that. I like to use the out of box alerts more so as a template, instead of actual production, due to the limitations on editing them.

If you are wanting to limit the 95%/Critical threshold CPU to a specific subset of systems, you can use "The scope of the alert: Only following set of objects" within the trigger condition of the alert. If you go that route, you would probably want to use a custom property or limit based off a naming standarding (Node Caption starts with X). Trying to maintain a list of systems within the alert becomes a management nightmare.