Re: [WISPA] Mikrotik Problem - 900Mhz-WDS-incremental speeddegradetp Zero then drop- repeat.
Update for those Interested I loaded the newest stable rel MT OS v 3.30 on all the radio. It did not help. The problem still existed. To review we had two clients a and b, and b was the one that would drop link if pass traffic in upload direction. Initially it was impossible to upload the firmware to the clientB. So I temporarily disabled clientA, and then it was possible to successfully upload new OSfirmware to clientB. So, I replicated the setup in the lab today, with 5 MT SBCs, of the same type as in the field. The only difference is I was out of XR900s so I used 5.Xghz cards. Initially I could not relicate the problem. So I decided to enduce some noise (a Trango AP randomly pointing to and away and to the test bed in a controlled fassion). I was able to replicate the problem. And yes the 411 system (equivellent to clientB) that had 5db better signal was the one that dropped link when the Trango noise was induced, just like in the field. What was most interesting is the results of the Bandwdith test, when noise was induced. Note we were simultaneously running 1500byte ping across both radios simultanous to MT bandwidth test to clientB, and accross clientA we ran a timed Iperf to generate triffic. . When noise was slowly induced, the pings stopped passing traffic first, then about a second or two later, the MT Bandwidth test (same results set at UDP or TCP) started the incremental slow down, 800mbps to 700mbps, to 500mbps, to 300 mbps until reached Zero, and then when at Zero the wifi session to ClientB dropped. So first thing we realized is that the MT Bandwdith test incremental slow down was a misleading symptom. Its the results the tool will always show when any Noise gets injected onto the link to the level that full packet traffic won't pass. Second thing noticed... In our original test bed, clientB was on Station WDS, and CLientA on WDS Slave. This is because clientB is the 411 board and has License level 3, and we figured it would only support station modes. We also switched ClientA to station WDS, and when we did that, and injected noise, it took a bit longer and more noise before the noise caused links to drop, and it also eventually caused ClientA to also drop along with ClientB. That last test was done at end of day, as we were finishing up. Tommorrow, we are going to substitute a 433board for teh 411 board, and see if we get different results or not. Tommorrow we are also going to try different configuration methods other than WDS modes, to see if the links drop as easilly in the same way or not. So in summary I can conclusively say The original way I had radio configured wa sperfectly acceptable for low noise conditions. But with 900Mhz, I surely will run into sporatic noise, atleast at that site.. It is clear that noise was integating the odd behavior from the MT radios. It is also clear noise was at the AP side. What we still will be investigating is how come one radio was effected more than the other, and if we can find alternate MT configs to allow clients to be more noise resilient. In a nutshell, disconnections occured to soon on the one unit. Tom DeReggi RapidDSL Wireless, Inc IntAirNet- Fixed Wireless Broadband - Original Message - From: Josh Luthman j...@imaginenetworksllc.com To: WISPA General List wireless@wispa.org Sent: Thursday, September 17, 2009 4:04 PM Subject: Re: [WISPA] Mikrotik Problem - 900Mhz-WDS-incremental speeddegradetp Zero then drop- repeat. WDS and nstreme can be used with wireless-test I hear. Before that it was not workable at all. Any load seems to kill your links - that has to be kept on mind. Josh Luthman Office: 937-552-2340 Direct: 937-552-2343 1100 Wayne St Suite 1337 Troy, OH 45373 When you have eliminated the impossible, that which remains, however improbable, must be the truth. --- Sir Arthur Conan Doyle On Thu, Sep 17, 2009 at 3:41 PM, Tom DeReggi wirelessn...@rapiddsl.netwrote: Well your problem reminded me of wds + nstreme problem is why I brought it up. I believe wireless-test will fix this. How can WDS and NStreme be used togeather? I thought it had to be one or the other? Any way you could test the links disconnected from the rest of the network and see if stressing the links drops it? Will do that if necessary, after firmware update. Are the links losing wireless association? Yes, they do when it reaches Zero mbps, then immediately restablishes association. Tom DeReggi RapidDSL Wireless, Inc IntAirNet- Fixed Wireless Broadband - Original Message - From: Josh Luthman j...@imaginenetworksllc.com To: WISPA General List wireless@wispa.org Sent: Wednesday, September 16, 2009 9:43 PM Subject: Re: [WISPA] Mikrotik Problem - 900Mhz-WDS-incremental speed degradetp Zero then drop- repeat. Well your problem reminded me of wds + nstreme problem is why I brought it up. I believe wireless-test will fix this. Any
Re: [WISPA] Mikrotik Problem - 900Mhz-WDS-incremental speeddegradetp Zero then drop- repeat.
Now that is very useful info Tom. I look forward to your next report. Thanks! -RickG On Wed, Sep 23, 2009 at 3:29 AM, Tom DeReggi wirelessn...@rapiddsl.net wrote: Update for those Interested I loaded the newest stable rel MT OS v 3.30 on all the radio. It did not help. The problem still existed. To review we had two clients a and b, and b was the one that would drop link if pass traffic in upload direction. Initially it was impossible to upload the firmware to the clientB. So I temporarily disabled clientA, and then it was possible to successfully upload new OSfirmware to clientB. So, I replicated the setup in the lab today, with 5 MT SBCs, of the same type as in the field. The only difference is I was out of XR900s so I used 5.Xghz cards. Initially I could not relicate the problem. So I decided to enduce some noise (a Trango AP randomly pointing to and away and to the test bed in a controlled fassion). I was able to replicate the problem. And yes the 411 system (equivellent to clientB) that had 5db better signal was the one that dropped link when the Trango noise was induced, just like in the field. What was most interesting is the results of the Bandwdith test, when noise was induced. Note we were simultaneously running 1500byte ping across both radios simultanous to MT bandwidth test to clientB, and accross clientA we ran a timed Iperf to generate triffic. . When noise was slowly induced, the pings stopped passing traffic first, then about a second or two later, the MT Bandwidth test (same results set at UDP or TCP) started the incremental slow down, 800mbps to 700mbps, to 500mbps, to 300 mbps until reached Zero, and then when at Zero the wifi session to ClientB dropped. So first thing we realized is that the MT Bandwdith test incremental slow down was a misleading symptom. Its the results the tool will always show when any Noise gets injected onto the link to the level that full packet traffic won't pass. Second thing noticed... In our original test bed, clientB was on Station WDS, and CLientA on WDS Slave. This is because clientB is the 411 board and has License level 3, and we figured it would only support station modes. We also switched ClientA to station WDS, and when we did that, and injected noise, it took a bit longer and more noise before the noise caused links to drop, and it also eventually caused ClientA to also drop along with ClientB. That last test was done at end of day, as we were finishing up. Tommorrow, we are going to substitute a 433board for teh 411 board, and see if we get different results or not. Tommorrow we are also going to try different configuration methods other than WDS modes, to see if the links drop as easilly in the same way or not. So in summary I can conclusively say The original way I had radio configured wa sperfectly acceptable for low noise conditions. But with 900Mhz, I surely will run into sporatic noise, atleast at that site.. It is clear that noise was integating the odd behavior from the MT radios. It is also clear noise was at the AP side. What we still will be investigating is how come one radio was effected more than the other, and if we can find alternate MT configs to allow clients to be more noise resilient. In a nutshell, disconnections occured to soon on the one unit. Tom DeReggi RapidDSL Wireless, Inc IntAirNet- Fixed Wireless Broadband - Original Message - From: Josh Luthman j...@imaginenetworksllc.com To: WISPA General List wireless@wispa.org Sent: Thursday, September 17, 2009 4:04 PM Subject: Re: [WISPA] Mikrotik Problem - 900Mhz-WDS-incremental speeddegradetp Zero then drop- repeat. WDS and nstreme can be used with wireless-test I hear. Before that it was not workable at all. Any load seems to kill your links - that has to be kept on mind. Josh Luthman Office: 937-552-2340 Direct: 937-552-2343 1100 Wayne St Suite 1337 Troy, OH 45373 When you have eliminated the impossible, that which remains, however improbable, must be the truth. --- Sir Arthur Conan Doyle On Thu, Sep 17, 2009 at 3:41 PM, Tom DeReggi wirelessn...@rapiddsl.netwrote: Well your problem reminded me of wds + nstreme problem is why I brought it up. I believe wireless-test will fix this. How can WDS and NStreme be used togeather? I thought it had to be one or the other? Any way you could test the links disconnected from the rest of the network and see if stressing the links drops it? Will do that if necessary, after firmware update. Are the links losing wireless association? Yes, they do when it reaches Zero mbps, then immediately restablishes association. Tom DeReggi RapidDSL Wireless, Inc IntAirNet- Fixed Wireless Broadband - Original Message - From: Josh Luthman j...@imaginenetworksllc.com To: WISPA General List wireless@wispa.org Sent: Wednesday, September 16, 2009 9:43 PM Subject: Re: [WISPA] Mikrotik Problem - 900Mhz-WDS