From: Peter van Oene (pvo@xxxxxxxxxxxx)
Date: Wed May 01 2002 - 17:13:10 GMT-3
The best thing to do is not send full tables to routers with 128 meg :) As
numerous other processes share the same memory, you are placing yourself at
risk. However, if you must, I would set the prefix max appropriately and
allow it to be torn down when it became exceeded. Simply ignoring updates
beyond the threshold might produce some weird behavior and would involve a
lot more coding that simply popping out a warning or dropping the
peer. One philosophy for protocol design suggests (as I understand it)
that in the event that strange things happen, it is much better for routers
to fully break, rather than partially break and potentially exacerbate
issues. In this case, dropping the peering session fully would serve to
protect the rest of the AS from a seemingly broken update.
At 03:49 PM 5/1/2002 -0400, Lupi, Guy wrote:
>Ok, I'll bite, I assume your talking about the maximum prefix feature, but
>that would either shut down your bgp session or if configured to, send you a
>warning. Is that what you are referring to, or is there some way to limit
>the prefixes and just drop any new ones that exceed a threshold? Thanks.
>
> > -----Original Message-----
> > From: Peter van Oene [mailto:pvo@usermail.com]
> > Sent: Wednesday, May 01, 2002 2:33 PM
> > To: 'ccielab@groupstudy.com'
> > Subject: RE: BGP tables increasing significantly
> >
> >
> > prefix limits are your friend
> >
> > At 10:53 AM 5/1/2002 -0700, Andy Singh wrote:
> > >Routers running BGP with 128mb ram are crashing because of
> > this, including
> > >ours.
> > >
> > >-----Original Message-----
> > >From: John Mistichelli [mailto:jmistichelli@yahoo.com]
> > >Sent: Wednesday, May 01, 2002 9:29 AM
> > >To: Howard C. Berkowitz; 'ccielab@groupstudy.com'
> > >Subject: Re: BGP tables increasing significantly
> > >
> > >
> > > I noticed it this morning.
> > > "Howard C. Berkowitz" <hcb@gettcomm.com> wrote: At 10:04
> > AM -0400 5/1/02,
> > >Lupi, Guy wrote:
> > > >Has anyone else noticed that their bgp prefixes received from their
> > > >providers has gone up by about 3 to 4 thousand routes in
> > the last week or
> > > >so? Is there an "issue" somewhere on the net? Or is this
> > something that
> > > >happens periodically and I am just now noticing it?
> > >
> > >A sudden jump has been reported by several carriers on NANOG. Not
> > >everyone is seeing it, and people are trying to figure out the cause.
> > >Quite a number saw about a 10K increase about 6:30 AM (GMT-5) this
> > >morning (May 1).
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:58:48 GMT-3