4.5 Upgrade problems

Of two APs I upgraded via cnMaestro early this AM, both came back online with incorrect configurations.

The first AP wasn't so bad, it had only chosen to switch from auto-selection of the sector/smart antenna to forced sector.  I changed the config, hit save, and things were back to normal. This unit was upgraded from RC-61 to 4.5.

The second unit came back up in a state that stopped clients from connecting.  On the dashboard it was displaying a 50/50 traffic split and no encryption, while the configuration clearly had 75/25 chosen and wpa2 configured.  This unit was running 4.4.3 prior to the upgrade.  Simply rebooting the access point brought it back to the correct running config.  Screenshots from that one are below:

I have noticed a few issues with 4.5 as well.  On my ePmP 3000 sector after updating to 4.5 the option to choose 2.5 or 5 frames was not their, it was specified in the release notes, I opened a ticket with support and they are looking into it.

I also had the sector/smart antenna option on forced to sector as well, put on auto and saved and it was good to go.  I have recently went back to 4.4.3 because one issue I saw was that my RSSI values and SNR values dropped 6-8db on all my SM on that sector.


@DigitalMan2020 wrote:

..... On my ePmP 3000 sector after updating to 4.5 the option to choose 2.5 or 5 frames was not their, it was specified in the release notes, .....


We updated 2 sectors so far (one from 4.4.3, the other from 4.5-RC69) to see how things held. We are not seeing the 2.5ms frames option on either AP. 

We did not see a drop in SM RSSI levels. 


@CWB wrote:

@DigitalMan2020 wrote:

..... On my ePmP 3000 sector after updating to 4.5 the option to choose 2.5 or 5 frames was not their, it was specified in the release notes, .....


We are not seeing the 2.5ms frames option on either AP. 


2.5ms frame support for e3k has been delayed indefinitely.

That can't be right, this is an e3k AP running 4.5 release and  2.5ms is enabled.

We are not seeing the 2.5ms frames option on either AP. 

We beta tested 2.5 and 5.0 ms frames in the Beta phase. On the previous generation of 1000/2000 N gear, there was of coursre quite a difference between 2.5 and 5.0 ms latency. However, on AC gear - the latency is SO much lower to start with - and it seemed there was very little difference in latency between 5 and 2.5 frames, and yet a hit in performance.

My understanding is that 2.5ms is still a work in progress, but that the latency difference won't be quite as dramatic as was years ago in the 1000/2000 series - since 3000/300 latency is so good already.

1 Like

I also do not see any option to change frame size on any of the equipment that has been updated to 4.5.  Not that it matters much to me, but it certainly isn't there as a configurable option.

We've done two towers so far. On one of them (all 300 series SMs), one of the SMs factory reset itself during the firmware upgrade. Resulted in a truck roll to fix it. That is one out of 35 upgraded SMs.  One AP turned off the beam forming antenna in the upgrade. Minor issue there.

We also tested 2.5 ms frame w/ 4.5 and found that it seemed to cause some F200 SMs to start constantly reboot. You can enable/disable it via cnMaestro.

Cambium really should provide better clarity on 2.5 ms support. Here is paste from release notes:

Support for 2.5ms frame size
2.5 ms frame size allows for co-location (Synchronization) with ePMP series radios. Configure the
frame size for use when in TDD or TDD PTP mode.

The BSA is at least staying connected now. We have 1 AP with 24 SMs. But we're now seeing large fluctuations in uplink RSSI (over 25 dB). MCS and retransmit look pretty much the same as prior to turning it on, so may be a reporting issue. However, I'm not seeing any benefit from the BSA on this AP. May try adding to another AP to see if it's any different.

A few SMs that had poor uplink performance are working much better now, however, one is not. Uplink MCS and retrans rate look great but can only get a few Mbps upload and the customer has a problem keeping their VPN & RDP session up. Move them to a poorer quality connection on a 2000 AP and they work fine.

I apologize for the mess with 2.5 ms frame support in the release notes. 2.5 ms frame support is not officially introduced in the 4.5 because of stability issues, however it was available for configuration in 4.5 Beta firmware. The plan is to introduce it in the next firmware release.

The default value for Uplink Antenna was changed to Forced Sector for ePMP 3000, please enable Auto mode if needed.

Im sorry for possible inconveniences and lack of information.

Tgank you.

1 Like

@3-dBnetworks wrote:

We also tested 2.5 ms frame w/ 4.5 and found that it seemed to cause some F200 SMs to start constantly reboot. You can enable/disable it via cnMaestro.

Cambium really should provide better clarity on 2.5 ms support. Here is paste from release notes:

Support for 2.5ms frame size
2.5 ms frame size allows for co-location (Synchronization) with ePMP series radios. Configure the
frame size for use when in TDD or TDD PTP mode.


So the realease notes that are currently shown are from the January RC54 build. They're not accurate. Someone at Cambium just copy/pasted the old beta release notes. There are things missing or inaccurate... like the 2.5ms frame support for e3k... that didn't make the final cut. Also, ACS was added to the e3k... which wasn't included in the release notes. There were upload fixes that were supposedly added that aren't part of those release notes. I've brought this to Cambium's attention, but they have not updated the release notes yet for some reason.

Thanks Eric, I thought it looked that same as RC54 but didn't check. But just to clarify 2.5ms, my understanding from Sakid's earlier post on the beta forum is that 2.5ms is still supported in 4.5 on SMs for backwards compatibility with 1000/2000 APs running 2.5ms. We have that running on a few force 300SMs now and were starting to add more. Hope I don't have to swap back for 200's. 

There is a topic in the ‘Ideas’ section for everyone to Vote on. The poster says the gains in latency AND in throughput for him are substantial.

1 Like