Today 85.119.176.0/21 was announced by AS20912 with 177 prepends. I noticed 20912 modulo 256 is 176. AS47868 modulo 256 is 252 which matches this mondays prepend-incident. So, what router OS will put 20912 into a byte and thus end up with 176 in something like "set as-path prepend last-as <no of prepends>" ? It needs to be fixed. Has anyone noticed any ill effects with IOS and using "bgp max-as"? Will it just drop any prefixes with long as-paths and no other ill operational effects? -- Mikael Abrahamsson email: swmike@swm.pp.se
Just seen that here too: Feb 19 16:20:35: %BGP-6-ASPATH: Long AS path 8001 8928 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 received from 207.99.64.25: More than configured MAXAS-LIMIT Our AS path limit is 100 which is way too high in my opinion but regardless I was trying to figure out any logic in this.... I can remember prepending one of our upstreams 4X at one point thinking that was a bit nuts .... thankfully we don't prepend anyone these days.... Paul -----Original Message----- From: Mikael Abrahamsson [mailto:swmike@swm.pp.se] Sent: Thursday, February 19, 2009 4:21 PM To: nanog@nanog.org Subject: lots of prepends Today 85.119.176.0/21 was announced by AS20912 with 177 prepends. I noticed 20912 modulo 256 is 176. AS47868 modulo 256 is 252 which matches this mondays prepend-incident. So, what router OS will put 20912 into a byte and thus end up with 176 in something like "set as-path prepend last-as <no of prepends>" ? It needs to be fixed. Has anyone noticed any ill effects with IOS and using "bgp max-as"? Will it just drop any prefixes with long as-paths and no other ill operational effects? -- Mikael Abrahamsson email: swmike@swm.pp.se ---------------------------------------------------------------------------- "The information transmitted is intended only for the person or entity to which it is addressed and contains confidential and/or privileged material. If you received this in error, please contact the sender immediately and then destroy this transmission, including all attachments, without copying, distributing or disclosing same. Thank you."
Mikael Abrahamsson wrote:
Today 85.119.176.0/21 was announced by AS20912 with 177 prepends. I noticed 20912 modulo 256 is 176. AS47868 modulo 256 is 252 which matches this mondays prepend-incident.
So, what router OS will put 20912 into a byte and thus end up with 176 in something like "set as-path prepend last-as <no of prepends>" ? It needs to be fixed.
Has anyone noticed any ill effects with IOS and using "bgp max-as"? Will it just drop any prefixes with long as-paths and no other ill operational effects?
No ill effects here, but I never saw the others before this one, and I'm only seeing it via 3561. 010308: Feb 19 13:08:13.455 PDT: %BGP-6-ASPATH: Long AS path 3561 3257 8928 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 received from 216.88.158.93: More than configured MAXAS-LIMIT ~Seth
The only ill effect is if set it too low.... we tested it a bit at 20-30 AS path length range figuring we shouldn't see *much* and it was staggering over time. The unfortunate thing more related to your question is that we found some AS's that were prepending 40-50 times to ALL their upstreams so with max-as set too low we had no routing to them at all! We've had it set to 100 for quite a while now and no side effects.... Paul -----Original Message----- From: Seth Mattinen [mailto:sethm@rollernet.us] Sent: Thursday, February 19, 2009 4:50 PM To: nanog@nanog.org Subject: Re: lots of prepends Mikael Abrahamsson wrote:
Today 85.119.176.0/21 was announced by AS20912 with 177 prepends. I noticed 20912 modulo 256 is 176. AS47868 modulo 256 is 252 which matches this mondays prepend-incident.
So, what router OS will put 20912 into a byte and thus end up with 176 in something like "set as-path prepend last-as <no of prepends>" ? It needs to be fixed.
Has anyone noticed any ill effects with IOS and using "bgp max-as"? Will it just drop any prefixes with long as-paths and no other ill operational effects?
No ill effects here, but I never saw the others before this one, and I'm only seeing it via 3561. 010308: Feb 19 13:08:13.455 PDT: %BGP-6-ASPATH: Long AS path 3561 3257 8928 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 received from 216.88.158.93: More than configured MAXAS-LIMIT ~Seth ---------------------------------------------------------------------------- "The information transmitted is intended only for the person or entity to which it is addressed and contains confidential and/or privileged material. If you received this in error, please contact the sender immediately and then destroy this transmission, including all attachments, without copying, distributing or disclosing same. Thank you."
Hi all, I am writing on behalf of AS8928. We have changed our BGP policy against AS 20912 to allow maximum of 20 AS prepends. Our NOC will communicate this issue to customer and when I will have some news why this happened I will update NANOG list. Best Regards Tomas Caslavsky +---------------------------------------------------------------+ + Principal IP engineer + + Interoute CZECH + + Nad Elektrarnou 1428/47 + + 106 00 Praha 10 + + Prague + + Czech RepubliC + + Direct Phone: +420 225 352 675 + + Mobile Phone: +420 731 492 872 + + Email: Tomas.Caslavsky@interoute.com + +---------------------------------------------------------------+ "the impossible we can do - miracles take a little longer!" "/earth is 98% full... please delete anyone you can." Paul Stewart wrote:
The only ill effect is if set it too low.... we tested it a bit at 20-30 AS path length range figuring we shouldn't see *much* and it was staggering over time. The unfortunate thing more related to your question is that we found some AS's that were prepending 40-50 times to ALL their upstreams so with max-as set too low we had no routing to them at all!
We've had it set to 100 for quite a while now and no side effects....
Paul
-----Original Message----- From: Seth Mattinen [mailto:sethm@rollernet.us] Sent: Thursday, February 19, 2009 4:50 PM To: nanog@nanog.org Subject: Re: lots of prepends
Mikael Abrahamsson wrote:
Today 85.119.176.0/21 was announced by AS20912 with 177 prepends. I noticed 20912 modulo 256 is 176. AS47868 modulo 256 is 252 which matches this mondays prepend-incident.
So, what router OS will put 20912 into a byte and thus end up with 176 in something like "set as-path prepend last-as <no of prepends>" ? It needs to be fixed.
Has anyone noticed any ill effects with IOS and using "bgp max-as"? Will it just drop any prefixes with long as-paths and no other ill operational effects?
No ill effects here, but I never saw the others before this one, and I'm only seeing it via 3561.
010308: Feb 19 13:08:13.455 PDT: %BGP-6-ASPATH: Long AS path 3561 3257 8928 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 20912 received from 216.88.158.93: More than configured MAXAS-LIMIT
~Seth
----------------------------------------------------------------------------
"The information transmitted is intended only for the person or entity to which it is addressed and contains confidential and/or privileged material. If you received this in error, please contact the sender immediately and then destroy this transmission, including all attachments, without copying, distributing or disclosing same. Thank you."
The only ill effect is if set it too low.... we tested it a bit at 20-30 AS path length range figuring we shouldn't see *much* and it was staggering over time. The unfortunate thing more related to your question is that we found some AS's that were prepending 40-50 times to ALL their upstreams so with max-as set too low we had no routing to them at all!
aha! an idiot filter. this could be a feature, not a bug. randy
participants (5)
-
Mikael Abrahamsson -
Paul Stewart -
Randy Bush -
Seth Mattinen -
Tomas Caslavsky