Re: BGP Announcement spike at LINX

[Erik Romijn <eromijn () ripe net>]

On Sat, Nov 18, 2006 at 04:43:47PM -0500, NetSecGuy wrote:
> I'm seeing a spike in announcements at LINX in the BGP updates from
> RIPE's RRC01 data.  It started around midnight UTC.  Announcements
> went from very low to 2K+ and have stayed there.  I'd like to be able
> to pinpoint what is causing the increase in announcements but it's not
> clear to me what the issue is.

One of our peers sending a full table started flapping every 1-2
minutes.

> I'm not seeing the same thing in the Routeviews LINX BGP updates.  Any 
> ideas?

Most likely only our peering flapped, as we use hold times of 0 and
probably few others do. We know of some routers that do not accept
this, causing flapping.
However, I haven't seen this occur suddenly with a session that worked
properly before. I've asked them whether they made any changes and set
this peering to use a normal hold time. It has been stable for 20
minutes now.

Yes, setting hold times to 0 has disadvantages of which this is only
one, so we are working towards setting this to normal values.

regards,
-- 
Erik Romijn                 RIPE NCC software engineer
http://www.ripe.net/        RIS operations

Attachment: _bin
Description: