15.1.3 Sustained Downlink Data Rate BUG

Ok, I need the communities help to help verify this bug for me.  I have been working with Cambium off and on for a few months now and they keep saying they can not reproduce it in the lab.  The good news is this bug is so easy to reproduce.  Any software version pass 15.0.3 on the AP side, the sustained downlink data rate is broke if you are using "Translation Bridging", as a note, no nat on sm and I have only confirmed this on multiple 450i aps, not sure about a 450 ap and also just 450 sms, I have not tested this issue on a 450i sm.  Here is a private link to a video of the issue, any comments, concerns, advice or help would be great.   

https://youtu.be/iZ7DO1TezOs

Thanks, Craig.

1 Like

Here are the config files...

Craig,

We were able to reproduce this issue in the lab this afternoon in light of your video.  We had been using a JDSU traffic generator for our testing which in theory should be equivalent to using iPerf.  Today we used iPerf as you did and were able to easily reproduce this.  We have collected data and are now working with engineering to work towards a solution.   We will keep you informed of the progress of getting a solution.

Chuck Hemesath

Director - Customer Support

2 Likes

Chuck, I would like to thank you for your response on this.  I would like to point out that I am a bit frustrated on how much time and effort I have went through to show your support engineers at different levels such an easy bug to reproduce.  I was even confirmed in the ticket that Iperf showed everything to work fine. This is from your engineers....

Test1.

  • Factory defaulted both AP and SM.
  • Upgraded radios from 15.0.3 to 15.1.2.
  • Connections: Laptop (iPerf server) --> AP <--> SM <-- Laptop (iPerf Client).
  • SM Max Throughput was UNLIMITED for the all the tests which we performed.
  • Performed iPerf test (which works as speed test but it will be between laptop to Laptop) with all default configuration and got the throughput about 60 Mbps.
  • QoS was set to Sustained DL/UL to 16 Mb/ 4Mb (rest all the values was 0) with Translation bridging enabled and disable and performed iperf test, so the results was as expected (Bandwidth was getting restricted). Note: Since we were testing in the lab we got bit low throughput.

I don't consider myself to know everything, but you might want to make it clear to your staff that when an operator with over 12 years of real world experience with Motorola/Cambium products the might want to take it more seriously.  I showed them over and over again, with webex sessions and did tones of testing, I was the one that even discovered it to be related to "translation bridging."   I first brought this issue up in ticket 123460  9 months ago, please review that ticket and you will notice that they tech said they fixed it in 15.1.1 and I asked for the tracking # on the fix because it was not in the release notes.  I than never tested the so called fixed and we continued to run 15.0.3.  We really needed to move to bigger channels and we have 450M coming so that is why this issue was brought up again when i started testing a few months ago.  So as you can see this has be a long and painful process that never should have needed to happen on a bug so easy to reproduce with an experienced operator.  I feel like it only finally was confirmed by your engineers when I finally decided to make this public.  I am will to provide anymore help or info that you need.  I just feel like you should know how this all when down, and how cambium for the first time in 12 years has let me down.  Thanks, Craig. 

1 Like

@craigschmadererwrote:

...

I feel like it only finally was confirmed by your engineers when I finally decided to make this public.  ...  Thanks, Craig. 


Yup - publicize how to turn a 4Mbps unit into an unlimited one and attention will be paid.  $$ talks.  


@CambiumChuckwrote:

Craig,

We were able to reproduce this issue in the lab this afternoon in light of your video.  We had been using a JDSU traffic generator for our testing which in theory should be equivalent to using iPerf.  Today we used iPerf as you did and were able to easily reproduce this.  We have collected data and are now working with engineering to work towards a solution.   We will keep you informed of the progress of getting a solution.

Chuck Hemesath

Director - Customer Support


With all respect - when you try to "simulate" a problem, well stuff happens.  It really can't be that difficult to set up a unit with "real" inernet traffic.  A simple speedtest would have revealed/confirmed the problem.  When we say we see something, it's because we see something, even if it's because we set up something incorrectly.  I know there are intermittent problems, but if you use the same config/setup we do - there should be a high probability of duplicating.  If you simulate, who knows?  Simulate = "Should" work the same.  Surely the Shuttle never "failed" during any of the simulations.

For the most part, I too have high regard for the engineering staff, but some times it's just frustrating trying to prove you aren't nuts and you see what you see.

2 Likes