Strange BGP announcement.

Brett Frankenberger brettf at netcom.com
Mon Nov 9 13:30:04 UTC 1998


:: Scott Gifford writes ::
> 
> > > > Bays don't crash (at least not in the general case ... for example,
> > > > mine stayed up this time and the last time this happened), but they do
> > > > send a NOTIFY and bring down the BGP session, as required by the RFC. 
> > > > (I believe gated does this also.)
> 
>   In case any Bay Networks users didn't already know this, reasonably new 
> version of the system software have a switch to turn off this behavior:
> 
> 1:TN]$g wfBgpPeerEntry.41.*
> wfBgpPeerEntry.wfBgpPeerASLoopDetect.157.130.101.182.157.130.101.181 = 2
> wfBgpPeerEntry.wfBgpPeerASLoopDetect.204.70.16.38.204.70.16.37 = 1
> wfBgpPeerEntry.wfBgpPeerASLoopDetect.204.70.100.66.204.70.100.65 = 1
> wfBgpPeerEntry.wfBgpPeerASLoopDetect.209.54.51.230.209.54.51.229 = 2
> wfBgpPeerEntry.wfBgpPeerASLoopDetect.209.54.101.238.209.54.101.237 = 2
> 
> [1:TN]$set wfBgpPeerEntry.wfBgpPeerASLoopDetect.204.70.100.66.204.70.100.65 2
>                           (41)                  (interface)
> [1:TN]$commit
> 
>   Set this flag to '2' for each interface to keep your router from tearing
> down BGP sessions when it finds a loop.  Don't forget to commit 
> afterwards, and then to do a "save config config" so it will take after 
> you reboot.  

This wasn't a loop.  This was a malformed AS path.  The length of the
entire AS Path attribute was 14 bytes, and the length of the first AS
Sequence segment was 7 AS's, even though there wasn't room to fit that
many in 14 bytes.  (7AS's * 2 bytes each + 2 bytes for the segment
header, gives a minimum of 16 bytes needed.)

Does this attribute also disable detection of malformed AS Paths?  

          - Brett  (brettf at netcom.com)
 
------------------------------------------------------------------------------
                               ... Coming soon to a      | Brett Frankenberger
.sig near you ... a Humorous Quote ...                   | brettf at netcom.com
 



More information about the NANOG mailing list