Strange BGP 'problem'
At the risk of posting something on topic.. I have 2 routers (lets call them A and B), talking to different upstream providers, advertising 202.6.75.0/24 via BGP. 202.6.75.0/24 is subnetted, each subnet is on a different VLAN. These routers talk to each other with iBGP, giving each other everything (including 202.6.75.0/24 le 32). I run HSRP between the 2 routers. Router B prepends my ASN several times before advertising 202.6.75.0. Relevant stuff (on both boxes): router bgp 23729 network 202.6.75.0 no synchronization ! ip route 202.6.75.0 255.255.255.0 Null0 With this config, reachability for 202.6.75.0/24 hosts alternates on and off roughly every minute. 47% total loss over 10 minutes (ie 5 'bursts' of loss). These packets are all hitting my AS at router A. I've turned HSRP off and on again to see if that was causing problems (it wasn't). Several solutions I've found: - Turning the iBGP session to router B off. (uhh..) - Turning syncronization on. My understanding of BGP synchronization, is that BGP will only re-advertise with eBGP prefixes learned via iBGP after learning them via another IGP if one is running. <snip> ec-br-1(config-router)#do sh run | inc router router bgp 23729 bgp router-id 202.6.75.1 </snip> I'm not running any IGPs, so synchronization shouldn't be part of the equation. In any case, I'm not a transit provider. And, there seems to be no relation to readvertising routes here, the packets AFAICT are getting to router A and being dropped. I've got no problem running running with synchronization on at the moment, but thats not the point. :-) Any ideas/thoughts? -- Nathan Ward
I would also be tempted to bump the administrative distance of this right up, to say 254. ip route 202.6.75.0 255.255.255.0 Null0 254 -----Original Message----- ip route 202.6.75.0 255.255.255.0 Null0
On 1 Nov 2003, at 16:56, Philip D'Ath wrote:
I would also be tempted to bump the administrative distance of this right up, to say 254.
ip route 202.6.75.0 255.255.255.0 Null0 254
That's entirely unnecessary, as long as there are more specific routes covered by that /24 in the RIB (and if there aren't, Null0 may well be a reasonable place to send the traffic).
Early IOSs had a problem with HSRP when ICMP redirects were enabled on the same interface. If your IOS is not new, try disabling ICMP redirect messages. This manifested itself as a loss of connectivity. interface a b/c no ip redirects standby priority ... standby preempt standby ip ... -----Original Message----- From: Nathan Ward [mailto:nznog(a)daork.net] Sent: 2 November 2003 1:36 a.m. To: nznog(a)list.waikato.ac.nz Subject: [nznog] Strange BGP 'problem' At the risk of posting something on topic.. I have 2 routers (lets call them A and B), talking to different upstream providers, advertising 202.6.75.0/24 via BGP. 202.6.75.0/24 is subnetted, each subnet is on a different VLAN. These routers talk to each other with iBGP, giving each other everything (including 202.6.75.0/24 le 32). I run HSRP between the 2 routers. Router B prepends my ASN several times before advertising 202.6.75.0. Relevant stuff (on both boxes): router bgp 23729 network 202.6.75.0 no synchronization ! ip route 202.6.75.0 255.255.255.0 Null0 With this config, reachability for 202.6.75.0/24 hosts alternates on and off roughly every minute. 47% total loss over 10 minutes (ie 5 'bursts' of loss). These packets are all hitting my AS at router A. I've turned HSRP off and on again to see if that was causing problems (it wasn't). Several solutions I've found: - Turning the iBGP session to router B off. (uhh..) - Turning syncronization on. My understanding of BGP synchronization, is that BGP will only re-advertise with eBGP prefixes learned via iBGP after learning them via another IGP if one is running. <snip> ec-br-1(config-router)#do sh run | inc router router bgp 23729 bgp router-id 202.6.75.1 </snip> I'm not running any IGPs, so synchronization shouldn't be part of the equation. In any case, I'm not a transit provider. And, there seems to be no relation to readvertising routes here, the packets AFAICT are getting to router A and being dropped. I've got no problem running running with synchronization on at the moment, but thats not the point. :-) Any ideas/thoughts? -- Nathan Ward _______________________________________________ NZNOG mailing list NZNOG(a)list.waikato.ac.nz http://list.waikato.ac.nz/mailman/listinfo/nznog
participants (3)
-
Joe Abley -
Nathan Ward -
Philip D'Ath