CPU Utilization

[lpopejoy]

I'm dealing with a lot of packet loss on a small network.  Support says that the CPU is hitting peak utiliization about 75% of the time.

Two questions:

1. There is abviously something going on that is causing the CPU to max, but I'm not sure how to isolate this issue.  What would you do?

2.  It is a bit frustration to have to call support and sit on hold for 40 min just to find out the the cpu utilization is.  Does anyone know if there is a way to get this data via API?   (note this is different that the performance score which IS available via API)

3.  Is there a way to get the built in chart for CPU Usage to work?  (org->Summary report->select a single network)

[ww]

Only support can see this cpu usage (not sure if they check it on a gui or cli)

What model you have? How many concurrent clients are connected?

Are you running 16.16 firmware? Some models  got performance improvements on 16.x

[cmr]

@lpopejoy the CPU graph appeared ages ago and we all thought that it would shortly be populated, unfortunately we are still waiting, even devices running 17.x don't show it... 

[lpopejoy]

Yeah...  I got excited when I saw it show up a while back.  

I even opened a support ticket today just to be sure this wasn't one of those magical "you have to contact support to get the feature enabled" things....  Here is their response:

Thank you for contacting Cisco Meraki support. My name is _ and I will be the support engineer assisting you on this case.
 
This is currently a known issue our development team is aware of. That graph is not populating because it is not a fully developed feature and is planned to be removed from the report due to this. Unfortunately, there is not any other information I can provide regarding the CPU utilization, but feel free to respond with the issues you are seeing on the device and we can hopefully troubleshoot it further.
 
Please feel free to let me know if you need any further assistance or have any questions.

[cmr]

Great; the update is to remove the graph as opposed to make it work... 🤦‍♂️

[BlakeRichardson]

@lpopejoy 

1. What model of MX do you have?

2. How many users?

3. If you disable Advanced Malware Protection does the situation improve?

4. Do you have any client VPN or site to site VPN's running? 

[lpopejoy]

1. What model of MX do you have?

       MX64

2. How many users?

       15

3. If you disable Advanced Malware Protection does the situation improve?

       Haven't tried, but I doubt it.  It is too erratic to test reliably.

4. Do you have any client VPN or site to site VPN's running? 

        No client VPN.  One other site via AutoVPN.

[BlakeRichardson]

@lpopejoy  I have seen advanced malware protection do funny things so it might be worth a shot. Also what version of firmware are you running? 

[lpopejoy]

16.16, but we have had the issue on other firmware versions at this site.

[QHoang]

We are having the same issue on 16.16, after running for a few weeks. We tried updating to 16.16.1 and still happening. Support says we're at 100% utilization. 

Support still hasn't offered an RMA but I reached out to our sales rep and they're sending us a "free trial" to troubleshoot.

On an MX450. 

[mb_dtc]

I'd be curious your experience with the mx450 from that, as all of ours are tipping over still after RMAing us from mx600's.  Using them in sdwan with upward of 1100 mx's in autotunnel to the hub(s), they can no longer keep up, particularly since 16.x last year, and support actually suggested they'd upgrade us first (then sales fought it, no commish in free!).  I adopted these for a customer after their last meraki guy ran away, and found out quick that 5000 max vpn count in the data sheet is a joke when things started breaking.  First the number given was "real world some ~1500" tunnels we heard from cisco, and the reality is they can't do much over 1100 now on the mx600/450 with them pooping out at some 85% usage.  Now faced with random outages they're having to shard all our branch networks out across 5 hub locations and soon a half mil $$$ later, the customer is not terribly happy.

Sad part is Cisco doesn't themselves have a good reference guide for sharding networks for very large branch quantities across hubs since the wilette.works site died...  Good thing I remember it to recreate.

[TheAlchemist]

I think device utilization in Summary Report is not the same as CPU utilization and as @ww already mentioned, only Support can see CPU utilization and it is not visible on Meraki Dashboard to users. Is there any API call which can provide such info?