09-08-2014 12:59 AM
We have plans to set up triggers for a few things to try and catch issues early on.
The triggers we have tested the last few months have been:
The values that I´ve set as a test were:
CPU Utilization => 75% for 10min
Memory Utilization => 70% for 5min
Bandwith Utilization => 100 Mbit for 10 min (AP 105)
The reason I set these values was to evaluate if we could gain something usefull from them. For example if a device covered a to dense area and needed to be off-loaded by another AP, thereby providing a better user experience or if the AP for example had hung or were experiencing any other issues.
Now, here are my questions:
CPU Utilization => 75% for 10min seems to happen on a frequest basis. Is it recommended to set a higher value? Say everything under 85% is fine and results in no user experience degration?
Memory Utilization => 70% for 5min - Seems fine as it has never been activated and the working AP:s are genereally well below the threshold.
Bandwith Utilization => 100 Mbit for 10 min (AP-105). This could probobly be set to a higher value. But if we speculate that there will be max 20 clients connected to one AP, could the recommended value be set to, if any, for bandwith uthilization? I know there may not be a simple answer to this. Just wanna get some input on the values :)
Solved! Go to Solution.
09-08-2014 03:54 AM
You may need to tweak these depending on your environment, but basically for a customer I have the following.
Memory Util >= 75% for 15 mins. The 650 controller and the IAP93 clusters are always above this, so we just ignore alerts for those. The IAP105 clusters and other 3000 controllers don't hit this threshold though.
CPU Util >= 75% for 15 mins. Generally this does not get triggered, except for us we get a flood of alerts for a particular controller every month. It turns out this is due to their vulnerability scanner hitting the device every month and going crazy over on of the OpenSSL vulnerabilities.
The bandwidth utilisation and clients per AP will depend on your environment though. I normally set the bandwidth per AP to be 40 Mbps.
If my post is helpful please give kudos, or mark as solved if it answers your post.
ACCP, ACMP, ACMX #294
09-08-2014 05:27 AM
You can also setup a trigger based on AP capacity as in number of clients associated to a radio over X time.
For the CPU, is that getting triggered in multiple APs? Can you post a CPU util chart over time? Is it pretty consistent?
Consulting Systems Engineer - ACCX, ACDX, ACMX
If you found my post helpful, please give kudos
09-17-2014 01:54 AM
I´ve been away and forgot to give you guys a thanks for th input.
I´ve adjusted the triggers a bit and will continue to monitor certain triggers to find a good fit for the enviroment they are in.