03-15-2019 02:41 PM
Hello,
Was really appreciate if someone can provide some assistances. We have a number of sites setup like the attached picture, all connect to AS65002 who is our MPLS provider.
Each site has an internet connections and a connection to our MPLS provider, over the internet we have a tunnel where we peer with eBGP to the other sites, we also peer with our MPLS provider via eBGP. Within each site we have iBGP relationship between the routers who peer with eBGP and all devices have IGP (OSPF) between them.
We have noticed strange behaviour (routes-flapping and routing loops between iBGP neighbours) across the network. We have networks behind Routers C on both sites which are passed into OSPF and networks from Site A which are reached via Static.
The network was setup where in Site A, the static route from Router A was redistributed into BGP and OSPF and Router B had redistribution from OSPF to BGP and vice versa . For Site B, both Router A and B had redistribution from OSPF to BGP and vice versa.
Routes which are redistributed from BGP into OSPF have a metric set of 1000 and type E1 on Router A at both sites, compared to Router B where they are set to a metric to 1 and type E1.
We want to make sure the path via AS65002 is always preferred as it’s a much faster connections, if the 100Mbps link fails the network would transverse over the internet, using the eBGP tunnel.
I have read previously that you should not redistribute your BGP table into IGP. If I follow the advise and when the 100Mbps link fails, as I’ve not redistributed BGP into OSPF on Router A on both sites, Router C does not know about the networks which lives on the other site and therefore drops the connections – is this correct?
Which routers should be redistributing to what routing protocol?
I guess we also need to filter routes from the iBGP and eBGP neighbours to avoid loops? One of the think we have seen is a route shows up from BGP and then OSPF and then vice versa, causing a bouncing loop.
Thanks in advance.
04-04-2019 04:35 AM
Sorry Paul, I edited my post and then noticed you replied to the original post.
04-04-2019 08:11 AM
Hello
No worries - has this cleared things up for you?
03-19-2019 09:31 AM
Hi John,
As Paul mentioned the loop is caused by redistribution of bgp to ospf at Router A and from ospf to bgp at Router B. When you did redistribute at Router B, you were redistributing all OSPF routes (internal, external 1 & 2). So, one way to solve it (I prefer this way) is to use TAGs as suggested by Paul.
An alternative way to handle this is to redistribute BGP routes to OSPF as E1 while redistributing all STATIC routes to OSPF as E2 (the default), and redistribute OSPF internal and external 2 to BGP.
Example:-
router bgp 65111
redistribute ospf 1 metric 10 match internal external 2
HTH,
Meheretab
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide