Hi,
We have some PMP450 that have values high and sometimes they have 100% CPU usage. Could this affect to the service that our customers receive? How can I decrease this CPU usage.
This happens with 16.0.0.1 version and 15.2 version.
Regards,
We've noticed this, but it's on older/orginal 450 AP's that have a lot of subscribers... like 50+. Is this what you're seeing as well? We haven't noticed it impacting customers yet.
Yes, it´s the same. They are older/original PMP450 with more than 50 customers. I´ve received some complaints and I didn´t sure if this CPU usage could be affecting.
Regards,
@franco87 wrote:Yes, it´s the same. They are older/original PMP450 with more than 50 customers. I´ve received some complaints and I didn´t sure if this CPU usage could be affecting.
Regards,
franco87,
Yes, unfortunately the original 450 AP CPU can get maxed out in this case. This is why we developed the 450i, which has a much faster processor. I recommend planning to upgrade to 450i or the 450m (if available in your frequency band).
It is possible where CPU is so busy that it cannot fill the frame. The smaller your average packet size, the faster this will happen. So if this is the case, the CPU might not be putting the maximum number of packets over the RF as possible. This would show up as reduced throughput and Frame Utilization, so that is one way to tell.
In the meantime, to decrease CPU utilization on AP there's not a whole lot you can do. Keep web auto refresh to disabled, and keep monitoring of the device to a minimum. Or of course move SMs to another AP.
Charlie
1 Like
Ok. Thank you. I will do.
Regards,
We too have been experiencing 100% CPU usage over the last week or so. It's been happening to 450, 450i, and 450M, with very little RF throughput on the units.
We have an open ticket with support to isolate the problem, as usage/modulations do not seem to be the source at all.
We also haven't upgraded the network to 16.0.0.1 yet (running 15.1.5 mostly), so if you're seeing the same issue (100% usage, no/very little throughput), I would assume it isn't firmware related.
We have disabled cnMaestro on one of our 450M's to see if there is a recent update or something similar that may be causing a polling loop, and have yet to see the problem again on that AP.
Generally, a reboot has "solved" the problem, but we have yet to isolate a cause.
Would be nice if their was a way to run "top" on the units to see whats chewing through cpu in instances like this
Sooo... this is happening more often, and there are some people on FaceBook that have noted this... it's now happening on 450/450i/450m AP's. Basically if the cnMaestro agent is enabled, after some time (sometimes days or weeks), CPU will reach 100% on the AP, and the AP will no longer communicate with cnMaestro. This has been reported for both cloud and onpremise versions using 16.0.1.
I hope Cambium is doing something about this as it's super annoying having to reboot AP's when this happens to get them to talk to cnMaestro again.
Hi Eric,
This issue is fixed in upcoming releases 16.1 and 16.0.1.1.
Could you please test with 16.1 beta
Thanks,
Subi
2 Likes
As noted in the Beta forum... we really want folks beating on the 16.1 BETA release that's out there. There are many many fixes and some cool new features that we'd like feedback on (Auto Contention Slots, anyone??)...
Thanks!
1 Like
I have quite a few AP's that have CPU's stuck on 100% while I dont think its causing problems for customers, i'm losing monitoring data. I was wondering if there is anyway to fix this via CLI without a reboot.
Any help would be appreciated!
Jason
- CPU Uti.PNG (76.2 KB)
Hi Jason,
Are you using cnMaestro ? and which build you are one ?
Could you please collect engineering.cgi logs and share the same at subi.s@cambiumnetworks.com
Thanks,
Subi
We are using Maestro version 2.2.0-r60 on-premise, would you like engineering.cgi from the AP's that are having this issue?