1.3.0 Firmware upgrade results


#1

For those that have not upgraded firmware yet, wanted to report my success story.

1.68 mile link using 2x40 in DFS. Was running 1.2.3 and upgraded to 1.3 I use PRTG to monitor both ping times and SNMP data of both sides of the link. Without changing anything other than firmware:

  • Ping times across the link went from 11ms prior to upgrade to 3ms after upgrade
  • PHY Tx/Rx went from 540/640 Mbps to 720/790 Mbps (as seen from AP)
  • SNR is finally reported (I was in the distance bug of the old firmware)
  • MCS went from 7 tx / 8 rx to 8 tx / 9 rx (as seen from AP)

So far, so good! Really nice to get the latency down. I am manual frequency, auto gender and 4ms TDMA window.

Only odd thing, and I had this in 1.2.3 as well. It reports to me that it would use 5190 as the auto frequency. That is possibly the worst frequency it could use. For one, my Cambium AP is on that exact channel and there is a ton of clean air in the DFS channels. It only wants to use either 1x80 or 2x40 in channels outside of DFS and those all are being used and have very high signals in the spectrum analyzer.


#2

Hi Chadwick,

Thanks for bringing your Auto Everything observation to our attention. We would like to diagnose this for you, but need remote access. I will contact you privately to arrange for this.


#3

what does auto gender do?


#4

We upgraded the firmware early Saturday morning. Seemed to run yesterday but had to downgrade to the old firmware Sunday morning in order for it to hold a link. Could not get it to work on Auto Everything nor any combination of manual settings we tried. Would run for a few minutes and then drop the link. Looking at the manage interface, all stats went to hell about 2am. Now that they are back on old firmware, things have been stable.

One thing I noticed after installing the new firmware was that the on-board radio to radio speed test showed improved numbers. 400mbps one way, and 150Mbps the other. When I did a core router to core router TCP/IP speedtest over the link, I could only get around 30Mbps real throughput??

Does anyone else out there have experience using these with Mikrotik routers? Any compatibility isses? Look forward to any comments.


#5

I am also battling throughput issues I can’t explain since upgrading. Reported bandwidth is higher but doing Mikrotik to Mikrotik speed tests, I can’t break 60 Mbps on a 250 Mbps link. I’m also about to down grade… Glad I am not alone.


#6

Hi John,

The “Auto” option available in TDMA settings is used to turn on the new low latency Auto mode.


#7

We are investigating a few reports of odd Auto Everything behavior. Most reports about throughput and latency are positive. We would be happy to review specific cases. Chadwick and Michael, we will take a look at your specific issues and report findings directly. If anyone else finds similar issues, please contact us via chat on Monday and we will be happy to review each case.


#8

Update on the issues noted above…we did find a bug and removed 1.3.0 from the download site. We will be publishing version 1.3.1 with the fix shortly.


#9

Can you elaborate on this bug and how it manifests? We have two links live on 1.3.0 and I was about to upgrade our third when I noticed this this morning.

We had some strange issues when putting up one of the new links with 1.3.0 and I’m wondering if this might be part of it.


#10

I have since downgraded back to 1.2.6 until the new firmware is ready.


#11

I had a backhaul on 1.3.0 in Auto Everything stop passing traffic this morning. Any chance this is related? A reboot on the AP side fixed it.


#12

What’s the roadmap for 1.4?


#13

Here’s a list of problems and solutions based on your feedback:

Auto Everything would not select DFS channels on B5-Lite. Since B5-Lite has a signal channel, we expected that downtime required for DFS clearance would not be desirable. However, in cases where heavy interference makes DFS the only option, it would be desirable to change channels. We are going to change the Auto Everything code on B5-Lite such that it selects DFS channels, and Exclusions can be used to avoid DFS if desired. Additionally, B5/B5c code will be changed to allow DFS channel changes if they are the only option.

There were several confirmed reports that Auto Everything recommended or selected channels that had interference on one side of the link, but not the other. This was caused by delays with exchange of spectrum data between radios. We raised the priority of this task to prevent that going forward.

There were a couple of reports of missing spectrum, which we are still investigating. A reboot solves that problem as a short-term workaround, but we will root cause and fix this as well.

There is a vulnerability that could cause firmware corruption during a power loss or brown out, depending on some specific code upgrade/downgrade paths. The primary symptom is a red power LED which indicates a boot problem.

We will have more info and a new firmware version (1.3.1) out in the next day or so to address these.

There will likely be a 1.3.2 firmware version released a little later with some additional fixes, features and Tx power enhancements.


#14

Chris, we just reverted down to 1.2.3. Thank you for the update. We experienced some link stability issues when we upgraded either with selected channels or Auto Everything. Once the new version is released I will do some more investigation and if the issue persists, work with them to identify the issue as we are unique with a moving ferry as one of the points.


#15

I haven’t tried Auto-Everything, but I have had significant improvements to latency and throughput since 1.2.3. Also our target/actual RSSI are now within 1db of each other where they were 10db off when on 1.2.3. 2.5mi 2x40MHz link is running very stable around 500/800 and latency around 1ms.


#16

Hi Everyone,

Great news! The Auto Everything issues are fixed, and we were able to include the revised OOBE limit for U-NII-3 in 1.3.1.

I would especially encourage anyone who has upgraded a B5/B5c in the past to 1.3.0 to install and keep this version 1.3.1 to avoid the firmware corruption vulnerability.


#17

Can you tell us some real life numbers regarding the OOBE limit changes?


#18

Hi Keefe,

Here is the FCC Memorandum and Opinion Order that explains the change in detail:

It translates to an EIRP limit of 53 dBm (up from 40 dBm) between 5725 and 5850 MHz.
http://backhaul.help.mimosa.co/backhaul-faq-maximum-tx-power

That means you’ll have more than four times the power for use in longer distance links, or for overcoming SNR challenges.


#19

When you say a sneak preview, is this the finalized version which I can go away and put on my units or do I have to wait for the official release on the firmware page ?

Thanks.


#20

Hi Stephen,

It is safe to install this 1.3.1 beta (issues above are resolved). The release schedule has been revised to tomorrow instead of today so that we can get more radio time in beta. I may have another version to post here later today as well.