I'm finding some odd behavior on my new Big Iron that I can't quite
explain/fix/workaround.
First, I was finding that a nail down static route for BGP wasn't
distributing as expected:
telnet at myrouter#sh run | inc 204.132.129.0
ip prefix-list NO_TRANSIT_AS seq 20 permit 204.132.129.0/24
ip route 204.132.129.0 255.255.255.0 null0 16
network 204.132.129.0/24
telnet at myrouter#sh run | inc 63.211.239.0
ip prefix-list NO_TRANSIT_AS seq 10 permit 63.211.239.0/24
ip route 63.211.239.0 255.255.255.0 null0 16
network 63.211.239.0/24
telnet at myrouter#
So, you'd think that with the null0 16 and 'redistribute static' in the
BGP configuration, I'd be good to go, right? Wrong. And this is why:
telnet at myrouter#sh ip route 204.132.129.0
Total number of IP routes: 140000, avail: 0 (out of max 140000)
Destination NetMask Gateway Port
Cost Type
204.132.0.0 255.254.0.0 4.79.81.193 1/1
B
telnet at myrouter#show ip route 63.211.239.0
Total number of IP routes: 140000, avail: 0 (out of max 140000)
Destination NetMask Gateway Port
Cost Type
63.211.239.0 255.255.255.224 0.0.0.0 2/1 1
D
telnet at myrouter#
A bit of googling reveals that apparently there is no such thing as "ip
classless" that is needed for Foundrys. Great. So I fixed the above by
simply removing and re-adding the same config pieces, by copying and
pasting them back into the configuration with 'no' and then back as usual:
telnet at myrouter#conf t
telnet at myrouter(config)#no ip route 204.132.129.0 255.255.255.0 null0 16
telnet at myrouter(config)#router bgp
telnet at myrouter(config-bgp)#no network 204.132.129.0/24
telnet at myrouter(config-bgp)#no ip prefix-list NO_TRANSIT_AS seq 20
permit 204.132.129.0/24
telnet at myrouter(config-bgp)#ip route 204.132.129.0 255.255.255.0 null0 15
telnet at myrouter(config-bgp)#router bgp
telnet at myrouter(config-bgp)#network 204.132.129.0/24
telnet at myrouter(config-bgp)#ip prefix-list NO_TRANSIT_AS seq 20 permit
204.132.129.0/24
telnet at myrouter(config-bgp)#^Z
telnet at myrouter#wr mem
telnet at myrouter#sh ip ro 204.132.129.0
Total number of IP routes: 140000, avail: 0 (out of max 140000)
Destination NetMask Gateway Port
Cost Type
204.132.129.0 255.255.255.0 255.255.255.255 drop 15
S
yeah, I changed weight from a 16 to a 15 because I was getting desperate
and couldn't think of anything else. I'm pretty certain that wasn't the
fix.
Apologies if this has been covered before. In searching last three
months lists, I found one similiar (unanswered) question:
http://puck.nether.net/pipermail/foundry-nsp/2007-August/001081.html
This isn't really a BGP problem per-se. I'm used to being able to
override learned routes with host routes to test various paths. So I
tried and ran into the exact same wall of a static route not being
installed:
telnet at myrouter#show ip ro 67.161.150.168
Total number of IP routes: 140000, avail: 0 (out of max 140000)
Destination NetMask Gateway Port Cost
Type
67.161.128.0 255.255.224.0 4.79.81.193 1/1 B
telnet at myrouter#sh run | inc 67.161.150.168
ip route 67.161.150.168 255.255.255.255 65.116.241.21
telnet at myrouter#sh arp | inc 65.116.241.21
24 65.116.241.21 0090.693a.1d99 Dynamic 6 1/3
telnet at myrouter#ping 65.116.241.21
Sending 1, 16-byte ICMP Echo to 65.116.241.21, timeout 5000 msec, TTL 64
Type Control-c to abort
Reply from 65.116.241.21 : bytes=16 time=38ms TTL=57
Success rate is 100 percent (1/1), round-trip min/avg/max=38/38/38 ms.
telnet at myrouter#show ip route 65.116.241.21
Total number of IP routes: 140000, avail: 0 (out of max 140000)
Destination NetMask Gateway Port Cost
Type
65.112.0.0 255.240.0.0 4.79.81.193 1/1 B
Gateway: 4.79.81.193
CAM Entry Flag: 0000000000000000H
(hah, that's directly connected as the previous "show arp" demonstrates,
wth?)
telnet at myrouter#show ip route 65.116.241.21 255.255.255.255
Total number of IP routes: 140000, avail: 0 (out of max 140000)
Destination NetMask Gateway Port Cost
Type
0.0.0.0 0.0.0.0 4.79.81.193 1/1 1 S
0.0.0.0 0.0.0.0 *65.116.241.21 1/3 1 S
telnet at myrouter#traceroute 67.161.150.168 icmp numeric time 1
Type Control-c to abort
Tracing the route to IP node 67.161.150.168 from 1 to 30 hops
1 162 ms <1 ms 2 ms 4.79.81.193
2 1 ms <1 ms <1 ms 68.86.91.5
3 1 ms 1 ms 1 ms 68.86.91.6
4 2 ms 2 ms 2 ms 68.86.103.5
5 3 ms 3 ms 3 ms 68.86.103.2
6 * * * ?
7 17 ms 11 ms 11 ms 67.161.150.168
telnet at myrouter#sh ver
SW: Version 08.0.01jT53 Copyright (c) 1996-2004 Foundry Networks, Inc.
Compiled on Mar 05 2007 at 18:45:55 labeled as B2R08001j
(3847670 bytes) from Primary B2R08001j.bin
HW: BigIron 4000 Router, SYSIF version 21, Serial #: Non-exist
==========================================================================
SL 1: B4GMR4 M4 Management Module, SYSIF 2, M4, ACTIVE
Serial #: CH49040190
8192 KB BRAM, SMC version 1, BM version 21
512 KB PRAM(512K+0K) and 2048*8 CAM entries for DMA 0, version 0209
512 KB PRAM(512K+0K) and shared CAM entries for DMA 1, version 0209
==========================================================================
SL 2: B8GC Copper Switch Module, SYSIF 2
Serial #: Non-exist
2048 KB BRAM, SMC version 1, BM version 21
512 KB PRAM(512K+0K) and 2048*8 CAM entries for DMA 4, version 0209
512 KB PRAM(512K+0K) and shared CAM entries for DMA 5, version 0209
512 KB PRAM(512K+0K) and 2048*8 CAM entries for DMA 6, version 0209
512 KB PRAM(512K+0K) and shared CAM entries for DMA 7, version 0209
==========================================================================
SL 3: B8GC Copper Switch Module, SYSIF 2
Serial #: Non-exist
2048 KB BRAM, SMC version 1, BM version 21
512 KB PRAM(512K+0K) and 2048*8 CAM entries for DMA 8, version 0209
We're thinking of getting another 5 of these, but based on the intial
behavior of things that just mysteriously start working for no reason
(or never work as expected), I'm getting cold feet.
Comments?
