I have some questions and need some help… I need use two link for share traffic or make failover link. I try to draw simple diagram to show and explain.
and here is setting ;
On Router1
/ip address
add address=10.10.2.9/29 interface=ether5-5GhzLink network=10.10.2.8
add address=10.255.255.3 interface=loopback network=10.255.255.3
add address=10.10.2.5/30 interface=Ether6-11Ghz network=10.10.2.4
How can i share internet on both link to use both of them ? and is possible to use 5ghz link for backup ? if one link down other link should continue.
and further more did i made correct settings for ospf ?
I read somewhere in forum If I am not mislead OSPF is universal tool for path finding. You can not transmit data in both ways at the sime time. You should try different methods but OSPF do If one link fails then the transmission will go with one link. OSPF does not double bandwith .
Have you tried w/o setting cost on the interfaces? If both links are directly connected on both routers, cost should be the same for both link routes.
Post
/ip route print detail
/routing ospf route print
You need to ensure both links have the same cost, playing with interface costs on both routers if needed, so that both links have equal cost, and ECMP kicks in.
thanks pukkita,
I changed the cost to 10 both of them and now traffic goes both interface.. is this correct implementation ? should i change the gateway on router-2 or any extra setting ? and can you check the settings of ospf if correct
I changed the cost to 10 both of them and now traffic goes both interface.. is this correct implementation ?
Yes, that’s ECMP in action.
should i change the gateway on router-2 or any extra setting ?
No need to do anything for failover, test that for yourself by disabling any of radio link interfaces. Traffic should flow through the other. Routing takes care of that.
Hi pukkita,
I cant paste what you need to here cause there is more then 3200 route on list. I use redistribute-connected=as-type-1 and all pppoe ip’s also in route table. I think this is also problem.
I thought you were working on getting rid of that (from the other thread on that subject)
In the meantime, if you want to cut down on the routing table, then you can filter your /32 announcements on the pppoe server by creating an aggregate route and filtering out any sub-prefixes of it.
If the PPPoE server’s pool makes assignments to users from 10.20.30.0/24, then do these steps:
/ip route add type=blackhole dst=10.20.30.0/24
/routing filter
add chain=ospf-out action=discard prefix=10.20.30.0/24 prefix-length=25-32
If you have no other rules in the ospf-out chain, then that should be good enough. If you have other filters, then you’ll need to make sure this rule comes before any kind of rule which would accept 10.20.30.x/25 .. 10.20.30.x/32
EDIT: One other note - I’m assuming that you’re also redistributing static routes from the PPPoE server, which is required for the blackhole route to be advertised. If you’re not advertising 10.20.30.0/24 into OSPF, then don’t activate the filter rule because that would break your users’ connectivity.
yes of course … after your advice from the other thread I rid of that thanks again.
/routing filter
add action=discard chain=ospf-out prefix=172.10.20.0/24 prefix-length=25-32
add action=accept chain=ospf-out
this is what im using now. but last one question in here. if pppoe_pool is not /24 and if it is /23 the prefix should change. am I true here ?
Yes after adding blackhole route im using redistributing static routes, but still cant rid from redistribute-connected=as-type-1 cause some special pppoe_ip adresess but routing table going to be cut down.
Mostly correct - you’ll also need to add whichever interfaces you actually do want to form adjacencies on and set passive=no on those. If you don’t do that, then your OSPF will drop when you add interface=all passive=yes
