[72896] in North American Network Operators' Group
RE: Next hop issues inside AS577 to AS852?
daemon@ATHENA.MIT.EDU (Mike Tancsa)
Sat Jul 31 14:17:25 2004
Date: Sat, 31 Jul 2004 14:21:31 -0400
To: nanog@nanog.org
From: Mike Tancsa <mike@sentex.net>
In-Reply-To: <C54A2385FF2CFD4A98B0B7D4FCD4483F16CBF2@exch2k.admin.cityne
t.net>
Errors-To: owner-nanog-outgoing@merit.edu
At 08:27 AM 31/07/2004, Krichbaum, Eric wrote:
>If it's every 60 seconds, I'd suspect the BGP timer is the root. They
>probably forgot to use next-hop self or a static route to a peer. The
>end result being that the route to the bgp peer is learned via bgp
>itself...
Maybe. Hard to say if thats what their default hold time is. I am still
seeing the odd hit in their network. For the sites we connect with inside
Bell, the tunnel LQR expire is 10 seconds and we have seen 2 big bounces
since routing around this morning. I emailed their noc, but no
response. The Bell looking glass doesnt seem to have any flap statistics
so I dont know if things are bouncing inside :(
It is looking different once again. Via Cogent to me,
194# traceroute 199.212.134.1
traceroute to 199.212.134.1 (199.212.134.1), 64 hops max, 44 byte packets
1 HSE-MTL-ppp12931.qc.sympatico.ca (209.226.183.241) 266.986
ms 229.511 ms 209.679 ms
2 Hamilton-ppp278329.sympatico.ca (206.172.130.250) 419.439 ms 219.614
ms 199.657 ms
3 kitcorr01-fe0-0-0.15.in.bellnexxia.net (206.47.229.8) 219.461
ms 239.605 ms 187.514 ms
4 badBellDNS (64.230.241.125) 221.605 ms 209.597 ms 199.655 ms
5 badBellDNS (64.230.242.194) 219.439 ms 227.155 ms 202.135 ms
6 core2-chicago23-pos10-0.in.bellnexxia.net (206.108.103.118) 229.386
ms 249.620 ms 346.174 ms
7 bx1-chicago23-pos11-0.in.bellnexxia.net (206.108.103.125) 228.434
ms 234.041 ms 219.645 ms
8 p13-0.core01.ord01.atlas.cogentco.com (154.54.11.29) 249.438
ms 239.589 ms 199.658 ms
9 p15-0.core02.ord01.atlas.cogentco.com (66.28.4.62) 239.441
ms 249.618 ms 229.679 ms
10 p5-0.core01.yyz01.atlas.cogentco.com (66.28.4.214) 249.409 ms 237.480
ms 231.808 ms
11 g0-1.na01.b011027-0.yyz01.atlas.cogentco.com (66.250.14.230) 251.504
ms 259.618 ms 219.669 ms
12 1572534Ontario.demarc.cogentco.com (38.112.5.166) 239.467 ms 249.593
ms 219.660 ms
13 tor-hespler-360-dslgate.sentex.ca (64.7.143.43) 229.489 ms 249.606
ms 214.908 ms
14 hespler-tor-360-i4.sentex.ca (64.7.143.46) 241.707 ms 229.615
ms 219.680 ms
15 ns.sentex.ca (199.212.134.1) 259.436 ms 239.613 ms 215.514 ms
Hops 6 and 8 were coming back as * * * on the traceroute a few hrs ago, but
packets were getting to and from me. Hopefully someone from Bell will pipe
up on or offlist as to what the problem was / is and if its resolved. Telus
is my main transit, and I dont like having to use such a blunt approach to
working around this issue :(
---Mike
>Eric Krichbaum, Chief Engineer
>MCSE, CCNP, CCDP, CCSP, CCIP
>
>-----Original Message-----
>From: owner-nanog@merit.edu [mailto:owner-nanog@merit.edu] On Behalf Of
>Mike Tancsa
>Sent: Saturday, July 31, 2004 7:52 AM
>To: nanog@nanog.org
>Subject: Next hop issues inside AS577 to AS852?
>
>
>Unfortunately, I am not a direct customer of AS577 otherwise I would
>open a ticket with them, but we have a lot of sites inside Bell Canada
>that need to reach us.
>
>Starting suspiciously at maintenance window time, we were seeing
>sporadic reachability issues coming at us from Bell. I am pretty sure
>its to us, and not the other way around as exiting out, I always prefer
>my GT/360 link and depending on the source IP it always works. The path
>back to me was via
>AS852 (telus) but I had to massively prepend to force it via someone
>else to get things working.
>
>But here are 2 traceroutes from inside AS577 (Bell) back to me
>
>
>Traceroute a)
>194# traceroute -n 64.7.153.1
>traceroute to 64.7.153.1 (64.7.153.1), 64 hops max, 44 byte packets
> 1 209.226.183.241 266.863 ms 219.712 ms 199.588 ms
> 2 206.172.130.250 219.408 ms 219.700 ms 209.619 ms
> 3 206.47.229.8 239.433 ms 195.376 ms 193.984 ms
> 4 64.230.241.125 229.460 ms 199.675 ms 209.669 ms
> 5 64.230.242.150 209.415 ms 189.707 ms 199.627 ms
> 6 154.11.3.25 239.440 ms 219.684 ms 208.091 ms
> 7 154.11.6.17 241.001 ms 217.379 ms 231.976 ms
> 8 64.7.143.44 229.456 ms 199.648 ms 189.635 ms
> 9 64.7.143.45 224.784 ms 194.333 ms 189.678 ms 10 64.7.153.1
>229.431 ms 209.654 ms 209.672 ms
>
>traceroute b)
>194# traceroute -n 64.7.153.1
>traceroute to 64.7.153.56 (64.7.153.56), 64 hops max, 44 byte packets
> 1 209.226.183.241 278.871 ms 221.986 ms 189.687 ms
> 2 206.172.130.250 219.433 ms 229.680 ms 339.650 ms
> 3 206.47.229.19 239.447 ms 189.621 ms 195.621 ms
> 4 64.230.241.121 223.483 ms 209.663 ms 199.695 ms
> 5 64.230.242.97 329.445 ms 209.667 ms 229.666 ms
> 6 64.230.242.181 239.463 ms 196.798 ms 192.568 ms
> 7 * * *
> 8 * * *
>
>
>Every 60 seconds or so the path back to me inside AS577 would change
>back and forth between a) and b). I dont know what Hop 7 on b) is. It
>could be another peer to AS852 (Telus) or just another internal router
>at Bell (AS577). Suffice to say, when taking path b) packets never get
>back to me.
>
>To work around it, I had to prepend out my AS852 link so that Bell comes
>back at me via GT/360 or Cogent.
>
>Anyone from Bell or Telus around to clarify where the problem is?
>Sadly, this is a holiday long weekend here in Canada :( The wheels fell
>off around 4:30 AM EST.
>
> ---Mike
>
>
>
>--------------------------------------------------------------------
>Mike Tancsa, tel +1 519 651 3400
>Sentex Communications, mike@sentex.net
>Providing Internet since 1994 www.sentex.net
>Cambridge, Ontario Canada www.sentex.net/mike