The Juniper PR in question is actually 836197.
On Wed, Feb 6, 2013 at 10:22 AM, Matthew Petach <mpet...@netflight.com>wrote: > On Wed, Feb 6, 2013 at 5:10 AM, Jonathan Towne <jto...@slic.com> wrote: > > On Wed, Feb 06, 2013 at 07:57:06AM -0500, Alex Rubenstein scribbled: > > # The question should be more along the lines of, "why aren't you > multihomed in a way that would make a 30 minute outage (which is > inevitable) irrelevant to you? > > > > The fun part of this emergency maintenance in the northeast USA was that > even > > folks who are multihomed felt it: Level3 managed to do this in a way that > > kept BGP sessions up but killed the ability to actually pass traffic. > I'm not > > sure what they did that caused this, or whether anyone but northeast > folks > > were affected by it, but it sure was neat to be effectively blackholed > in and > > out of one of your provided circuits for a while. > > > I recommend you grab > http://kestrel3.netflight.com/2013.02.05-NANOG57-day2-afternoon-session.txt > > and search for PR8361907 > > Richard did a very good lightning talk about why > Juniper boxes will bring up BGP but blackhole > traffic for 30 minutes to over an hour, depending > on number of BGP sessions it is handling. > > His recommendation--if you don't like it, go tell > Juniper to fix that bug. > > Matt > > -- Brian C Landers http://www.packetslave.com/ CCIE #23115 (R&S + Security)
