I have for a long time now a BGP session established with ISP1. We have our own PI space which is for example 4.4.4.92/22
This is working fine.
Recently we had a second connection installed from ISP2 at the other end of our network and have a BGP session successfully established from them.
But…They do not seem to be advertising our PI space properly, or else its something I am doing wrong.
I have updated our route object in the Ripe Database with ISP2’s AS number.
Under BGP/NETWORKS I just slapped in 4.4.4.92/22 on the second router. Is this right?
On our second router I created a loopback address of 4.4.4.94.1, but when I do an external traceroute to 4.4.4.94.1 the trace just keeps tracing between ISP1 and our 1st router constantly.
Any advice greatly appreciated
BTW…how do I view the full BGP routing table on our second router?
I have just realised that we are only receiving the default route from one provider, and the full routing table from the second provider.
Could this be what is wrong perhaps?
If you have a full table from one provider and a default from the other provider, then you will probably see most of the traffic going out the connection with the full table.
Ok, We are receiving the full table from both ISP’s now.
But there is just one more issue wrecking my brain
We received a /22 public block from Ripe which has obviously 4 x /24’s in it.
I want to prepend all the blocks so as 2 x /24’s are preferred by ISP1 and the remaining 2 x /24’s are preferred by ISP2
I took out the 4.4.4.0/22 network statement out of BGP - NETWORKS and replaced it with the 4 x /24 blocks instead and created prepend rules.
Straight away the routers stopped announcing the public network range.
This is obviously the wrong way to do it.
The Public address range only gets advertised if I put back the 4.4.4.0/22 range in network statements under BGP.
Please help, as this is driving me crazy and there is no documentation out there that I can find to point me in the correct direction.
This will be based on the filters setup by your upstream provider if you haven’t set any filters on your own. You can advertise a /22 by splitting it into 4 /24.
Thanks for the reply,
So are you saying, that when I took the /22 out of BGP/NETWORKS and replaced it with just the 4 x /24’s, my routers should of announced them.
I also noticed that when I done this, the BGP Peer was very slow at updating their routes. They were populating, but a lot slower than normal.
wouldn’t you put all 4 /24’s into one AS, peer that with both ISP’s. then AS stuff so you get the path selection that you want.
So
1.0.1.0/24 … for preference on isp1, then AS pre append 2 for ISP2
1.0.2.0/24 … for preference on isp1, then AS pre append 2 for ISP2
1.0.3.0/24 … for preference on isp2, then AS pre append 2 for ISP1
1.0.4.0/24 … for preference on isp2, then AS pre append 2 for ISP1
Prepending is a valid way to do that but there is a little easier way to do it…just advertise the /24 prefixes you want to be preferred and then a /22 for failover. The reason for this is that prepending doesn’t always guarantee a pipe will be preferred - when you advertise specific subnets, the traffic will almost always come in exactly where you advertise it.
I know before I did strip the /22 from BGP NETWORKS, and just put in my 4 /24’s and it didn’t work. But I did not put in the /22 at the end for failover. I will try this tonight after 12am again.
Ok I have just confirmed it tonight…my peers do not like me advertising the separate /24’s out.
As soon as I remove the /22 prefix from bgp networks, and populate it with the 4 separate /24’s, the net stops and the peers stop advertising them out.
Depends on the type of filtering being used, but the answer to your question is yes…the upstream provider is only allowing a /22 mask through and you will have to call their NOC and request the ability to advertise /24 prefixes.
Please list the output of the following commands while you have the /24 networks advertised.
[admin@Falcor] > routing bgp advertisements print[admin@Falcor] > routing bgp network print
Hi Guys,
Since I last posted in this thread, our ISP2 went down for a few days.
On our BGP router facing ISP2, under BGP/PEERS our IBGP prefix count was 480k+ routes.
But we had a problem with the traffic not falling over to ISP1, and it involved me changing a lot of settings to try figure out what was wrong.
Since then, ISP2 came back, and now their prefix count under BGP/PEERS is showing 480k+ routes.
But…now my IBGP between both my AS routers is only showing a prefix count of 1.
Before both were almost matching, IBGP and ISP2.
The fact that I am only advertising out 1 prefix to my peers, could this be correct or should I be seeing 480k+ routes between my peers?
I am only using an EOIP tunnel between both my BGP routers as I could never figure out mpls.
