А також хотілось би знати чому я не можу налаштувати агрегацію по протоколу LACP під routerOS?
The epmp radios do not pass the lacp packets so the bridge will not come up.
We had a working hack around this but it no longer works. If you have a routerOS device on both ends just use OSPF to bond the links together and set the routing metric exactly the same to provide a pseudo load balancing. This will actually work better for you since even with LACP your maximum available throughput is still only what a single link can provide, but with routing you also get a much faster failover state in case one of the links drops but the radios are still up.
This is because LACP relies of the port link state to determine if the link is up and only sends confirmation packets every 5 to 10 minutes.
А можна показати як це сконфігурувать в мікротику. А то з LACP все просто, а от з трішки OSPF складніше
The funny thing about OSPF and its counter part ISIS is that the actual implementation on a router is very simple, its the concepts that drive one nuts!
You enable OSPF on the router and set the router ID. Then add the interfaces to the OSPF and set the area type all in one command. Done. Well almost, but this gets you running and you can tweak from there.
Here is a wiki to follow, its designed around a three router setup but you can just do two routers: Manual:OSPF-examples - MikroTik Wiki
Make sure your MTU on the path is large enough. Routers should be 1528 and all devices in between should be larger to make sure that the full packet can pass unfragmented. Normally OSPF packets will not be larger than 1500 but if your using vlans you will need the 28bytes.
As for now LACP functionality is not implemented. And we don’t expect it due to low demand.
I suggest to use some routing instead.
This is what I am steering him towards.
Andrii, can we split this off into its own thread to keep the original clean?
@Douglas_Generous, you are absolutely right!
I talked to Viacheslav and he has plain L2 network. Even equipment is in the same vlan with users. So now he has a piece of work to move it from L2 to L3.
From time time to time we get requests for this feature. But it used so rarely by some small ISPs. And it requires significant bridge code changes and big test plan. So it is in our road map with quite low priority. But we will do it someday!
By the way, maybe someone knows any cool trick to get aggregated link via radios without LACP and routing?
Telecommunication engineering sometimes look like real art! Sometimes like Picasso’s ))
As far as I have been able to determine, the only workaround that we had stopped working with the new firmwares. (thinking it was a hole in one of the packages that was “fixed” by the developer).
Not knocking a L2 network but they do get to a point of being problematic, dual paths being one of the larger problems.
As for a quick and dirty bypass that can be done, simply create one routed segment with two paths. Takes two routers and three ports on each. There is no rule that you cannot have the same IP space on both ends, just can’t have two devices using the same IP address.