Will this bonding failover work properly in a PtMP setup, when done only on the station side?
(Cube as station, connected to separate 5GHz and 60GHz sector APs on the same switch.)
Any chance a similar 5GHz backup option will be added to nRAY? It would be more useful at longer distances - at short distances up to 200m supported by wAP60/Cube, you can simply use LHG60/nRAY instead for extra rain fade margin. PtMP at 800m is too unstable with 60GHz alone, about 400-500m is the real max and somewhat higher gain 5GHz backup would be useful.
Any chances the new hardware (nRAY/Cube) will be capable to use the channels 5 and 6 (already allowed in the USA, proposed for EU as well)?
(66000 is non-standard and overlapping between 4 and 5, proper center frequencies are 66960 and 69120)
Bonding failover is waaaaaaaaaaaaay too slow. Modern day failover times should be less than a second
Only 1 way I know of to accomplish this in the MikroTik world, and thats with OSPF and BFD
The way we currently do it is to enable BFD on the primary link, but disable BFD on the backup. As BFD is not reliable on MikroTik so if you do have any sort of routing issue or adjacency flapping, at least traffic continues running over backup link and doesn’t flap on both (also failing over from high bandwidth 24/60ghz link to backup 5ghz link can massively congest it for a little while)
If there’s a better way i’d like to hear about it, but bonding failover times are what ~10 seconds? or more. I don’t think it has any fast L2 based failover
Would be nice if MikroTik Implemented ERPS
Yes, this can be reconfigured to work in such way. On 60GHz AP side add VLAN, that goes to 5GHz AP and configure as in wiki example, but replace 5GHz interface with VLAN.
It doesn’t work. Maybe I’m doing something wrong.
Could you please post example configuration with two Cube ac as stations with backup via 5GHz AP?
Thanks a lot
Care to elaborate further?
I suppose being the interfaces are directly on the mikrotik device it might be able to detect a link down event faster since the transmitting interface itself will go down and immediately be removed from the bonding interface (still takes time for it to drop though)
In my testing in a more typical setup (Router->AP->SM->Router) where the actual transmission may not be active but the interface is still up (since the ethernet doesn’t go down) it’s not fast at all and takes many seconds to detect the failure