Hello. We have 1,5km 2x2 P2P link with dual parabolic antenas. All works fine with NV2, both sides have 5.24
Troughput is 55Mbit. CCQ and link uptime is good.
But when I change protocol to Nstream, then periodic disconnect. Why?
i have the same problem, its kinda unstable, i use Nv2 with some clientes, its more stable but it have problems with the througput behind de wireless equipment on TCP protocol 
but Nstreme is stable on throughput :/, i think we have to wait for some fixes about the protocols.
BR
They dont even work on this issue
Typically, if I have a problem with nstreme it is usually related to local noise, another AP on my tower or something similar.
Nstreme is kind of unstable. We tried it on several dozen links in laboratory, but it behaves like a monkey with a granade in the hand. It drops thruput instantly for 70Mbit and then returns etc.. maybe it needs some re work, as it bring very good latency results..
This is happening because your data rate is drooping to lower speed (if u tested in lab that mean its not noise or interference problem) dont know why in my case most common it is happening when the link is on high load
when u do disable/enable on wlan interface it is going back to normal
U can fix that in data rate u can select best rate - but always its not working like it should be (not for me)
also I hate disconnect/reconnect problem
nstreme definitely need makeup
My feel is it is matter of speed set function fail. For example NV2 is made by small speed step ups , step downs in slow decissions, so it takes a time to get on operating speed quite some time, but then it is really robust. Nstreme is like to switch speed immediately, once it looses several pacakges and then it is very hesitative to get back on former speed (same as by AirMax), like it wont resend only demaged packages in lower speed.. it is like monkey on steroids.
Take speed select from NV2 and give it to Nstreme..
nstreme use to work fine on all links until nv2 was developed
I really wish they would improve the performance with Nstreme and 2X2. I use Mtik for most of my backhauls, and the latency has increased on each PtP to now 3-8ms. With good old Nstreme, it was a solid 1ms MOST of the time!!!
I have to use NV2 now, because I get more throughput, like 40% more. It would be nice if they could have an option to lower latency on NV2, in exchange for a little lower throughput, or fix Nstreme to work more effeciant with 2X2.
Any comments from Latvia?
if you have problems with Nstreme wireless protocol with lot of disconnection “too many poll timeouts” please write to support and also provide us remote access to both ends so we could check why it is disconnecting so frequency.
The problem could be that you are testing the link in the laboratory - the devices are too close to each other, the signal is too strong, you could also experience some misconfiguration. How are you measuring the throughput?
I can do that, but it is common problem of thruput stability met by the others as well. Try it your self , I am completely other stuff, we are operating polution free enviroment on 9GHz with your equipment, and it behaves diametrically other way then NV2, which stability is fine, even with speed rising, but during high rate transmitions (180Mbit thruput TCP + on radio) it behaves really strange. Espesially during traffic pattern changes. You should get same reusult once you will make line on your table.. I tested more than couple hundreds of units during last several week and it is all the same.
Simply not.. The feedbacks are from customers in real live testing as well, so there is some -65dBm on RX and range is in several hundreds m. Even in laboratory are radios some 2-3 meters away from each other. Power is set on 2 dBm, and there is 65dB attenuator on cable connections between them - it is not lab issue.
OK, I can get some pieces which are just wonderfull until next reset 
But this is not deal.
True, nstream no longer works ok. The last 15 days I have commissioned 10 new links based on the RB911 and JRC duplex 24 antennas. Signals on the links are -50 to -60 but none of the links with active NSTREAM has no uptime higher than 2h. Log on the client side says “lost connection, not polled for too long” and log the AP side “disconnected, too many poll timeouts”. Please note that the area is absolutely without interference. On RouterBOARD is OS version 6.1 and version 3.07 firmware. So, 10 links on 10 different places, interference-free area. With NV2 everything works ok but with big jitter, ping varies 4-50 ms.
I found out RouterOS 5.14 works stable with NSTREME on two chains,
where RouterOS 5.25 or 6.1 wont’t work stable in the same situation.
Tested on several real life production links of varying distance and rf-“crowdiness”.
Normis. You say - it is known bug. Why you don´t fix it?
http://forum.mikrotik.com/t/mikrotik-groove-as-a-ap/67369/2
It is old bug and very serious.
Thanks
I did not say it’s a known bug. it is a simple error message (poll timeout). it can indicate all kinds of things.
normis, but what do you say about my findings?
could you test if this problem effects AP, Client or Both.
Could you test in which cases you have problems?
- AP v5.14, Client v6.1
- AP v6.1, Client v5.14
- AP v6.1, Client v6.1
Then tell us in which of those three cases the nstreme is working fine and in which not.
After the findings maybe you could provide us with remote access to this link where you have the above problems.