Poll timeouts with ROS5.x / SXT

RouterOS General
1

Hi there,

I’m running an SXT to SXT link over about 2.5km. The link is great - all chains enabled with above channel extension (40MHz) we get sync of 216-300Mbps and about 100Mbps actual throughput on UDP bwtest. The link is running Nstreme with Polling and CSMA enabled, Policy = none, Framer limit = 3200.

Here’s the issue:
If we’re running ROS 4.17 the link is rock solid stable for weeks on end.
As soon as we load ROS5.0/5.1, the link keeps dropping… log says: “disconnected, too many poll timeouts”

If we disable nstreme, then eventually the link disconnects and stays disconnected: the bridge SXT shows the client is associated under the “Registration” tab, but no response to ping/mac ping. If you check from the client side, it’s not connected and it can’t scan the AP (bridge) either - the AP side needs reboot to fix)
NV2 does not run as well as nstreme for this link.

If the client side runs ROS5 it’s fine, but if we put 5.0 at the bridge side, it breaks

We have to stick to ROS4 for now in order for our link to work…

Any idea what would cause this difference in behaviour? I’ve had that error before with some of my other links, if the signal is really bad, but there’s nothing wrong with this link (CCQ is always above 80% and rates are always good… it just randomly keeps reconnecting with that error in ROS5)

Thanks

2

Bump.
Same problem with 5.2. The Bridge side needs to be running 4.17 then its super stable - 5.x causes disconnects constantly

3

Tried reset-configuration and rebuilt the link from scratch, same problem happened… It ran fine on nv2 for about 2 hours then logged the error “disconnected, not responding”, and the client could no longer scan the bridge - the bridge side needed reboot. Doesn’t happen on 4.17 so I find it difficult to see this as a hardware fault on the unit…?

4

Seems like the problem stopped since ROS 5.3, and now running ROS5.4 which is also going stable so far! Great job mikrotik & thank u