Here are some charts from one of our APs. It has 34 SMs registered. In the early morning hours of the 14th I made some changes. I upgraded from 15.1.1 to 15.1.4, changed frame time from 2.5ms to 5ms, and also increased contention slots from 3 to 4. It doesn't seem like throughput is quite as high as before (I was expecting more!), but how could the Frame Utilization be so much lower when it used to be topping out nightly at 100%?
PS I'm using Zabbix. There is some missing data on the 11th and the 12th. And the light/dark areas show when normal business hours are (~8am-6pm M-F for the light areas).
I'm assuming this is a 450 or 450i AP and not a 450m (my usual preferred AP for answering questions on) ?
5ms frame is inherently about 15-20% more efficient over the air compared to 2.5ms (due to turnaround time in tx<->rx gaps) so I would expected a drop in utilization for the same throughput when changing from 2.5->5ms.
This looks a bit too much to be just that though, are you sure all SMs came back to this same AP or might some of them have moved to a secondary choice ?
I think I've proven it seems to be an SNMP reporting issue, see captions below. PRTG is polling every 60 seconds, maybe something has changed with the calculation or reporting for that oid with the jump from 15.1.1 to 15.1.5 Maestro seems to be showing full or close to full frame utilization as expected.
I will decrease the PRG polling interval to see if that helps.
All SMs came back after upgrade, and that was the only AP I upgraded in that area. The upgrade was slow when I decided to turn on "Stop update On critical error" in Maestro 1.6.0 OnPrem.
At 5:45pm I changed a different 15.1.1 AP from 2.5ms to 5ms and the DL Frame Utilization via SNMP since then has been intermittent. No other changes were made, so it seems the SNMP Frame Utilization metic has an issue with 5ms frames.
You're right, the Frame Utilization software had a bug with 5 ms frame. This was fixed in 15.2 (Build BETA-4) and forward. (Internal issue # CPY-14774.)
It would be great if you would verify the fix on the beta, or wait till 15.2 goes official. FYI, 15.2 is planned to go official in a few weeks unless of course we discover other new issues in testing.
It does seem to be corrected. Around 4AM on the 11th I upgraded the firmware, and the RF Utliization appears believable again. I noticed the Radio traffic direction has flipped as well. Thanks for taking care of the issue. We'll probably wait until 15.2 to roll up to 5ms.