07/01/99 6Bone Routing Report

[Masaki Hirabaru]

We are still seeing these bogus routes.

  0000::/0 path 1225 33 109 237 5761 (MSR-REDMOND -- 0%)
  1000::/3 path 1225 33 109 237 5761 (MSR-REDMOND -- 0%)

Dorian Kim at as 1225 tried to catch them, but they are flapping
too rapidly. However, I found that CSELT also detected this on
their router according to their routing page.

I did an experiment to modify a route 2010::/16 originated from
5761. BGP attribute of the routes 0000::/0 and 1000::/3 followed
the change. There is another supporting fact. CAIRN was also
originating 2010::/16 before, and the same bogus routes from
CAIRN were also seen. After stopping 2010::/16 from CAIRN, the
bogus routes from CAIRN went away.  I'm sure that these bogus
routes never pass through AS 237 (Merit).

I have no idea about fixing these routes for the time being.  If
you are on or close to the above path, and find something that
might to be related to this, please report it.

BTW, I also tried to attach BGP community to the route
(2010::/16).  However, the route having traveled through 6bone
back to us didn't have BGP community. BGP community is optional
transitive, so an intermediate router may not know it but must
pass it to next.  Chris Heermann suggested me to use BGP
community to help 6bone routing, but it seems it's not possible
for the time being.

Masaki

>> From: Masaki Hirabaru <masaki@merit.edu>
>> Subject: Re: 07/01/99 6Bone Routing Report
>> Date: Mon, 19 Jul 1999 13:05:26 -0400
>> Message-ID: <19990719130526B.masaki@merit.edu>
>>
>> Hi. 6bone folks,
>> 
>> This issue looks like a matter local to some sites, but I'm
>> bringing up this because I'm not sure where it's generated.
>> 
>> Merit has been receiving 1000::/3 and 0000::/0 since July 1st.
>> They are out of the 6bone prefix, so they should not be. But, the
>> problem I want to say is not such a thing. I don't want to solve
>> this just by filtering out them. (Merit doesn't accept such a
>> route with an as path loop, anyway.)
>> 
>> The AS path of them is "1225 33 109 237 7081", where 7081 -
>> CAIRN, 237 - Merit, 109 - CISCO, 33 - DEC-CA, 1225 - CICNET.
>> 
>> As long as I checked over the BGP session logs that record all
>> BGP packets received here, Merit (AS 237) didn't have the routes
>> from CAIRN (AS 7081). According to our routing daemon's internal
>> log, it didn't announce the routes to CISCO (AS 109). However, it
>> received the routes as if they were originated from CAIRN through
>> Merit.
>> 
>> Moreover, this routes are flapping. They are only alive for a
>> second. Withdrawals follow right after the announcements that
>> happen once every 10 or 20 minutes.
>> 
>> This doesn't seem to cause something wrong for now, but I'd let
>> you know what's being seen.
>> 
>> BTW, there are also similar flapping routes that look from WIDE
>> Project and contribute great increase of IPv6 traffic on 6bone.
>> 
>>   0000::/0 path 109 2500 2500 2500 2500 2500 (WIDE)
>>   1800::/4 path 109 2500 2500 2500 2500 2500 (WIDE)
>> 1.  (0000::/0) had 93371 BGP+ updates (18 unique aspaths)
>> 2.  (1800::/4) had 87912 BGP+ updates (16 unique aspaths)
>> 
>> Thanks,
>> Masaki
>> 
>> >> From: Masaki Hirabaru <masaki@merit.edu>
>> >> Subject: Re: 07/01/99 6Bone Routing Report
>> >> Date: Mon, 19 Jul 1999 10:41:52 -0400
>> >> Message-ID: <19990719104152B.masaki@merit.edu>
>> >>
>> >> I restarted about 45 hours ago, but the same thing is happening.
>> >> I'm going to bring up this issue to 6bone mailing-list. -- Masaki
>> >> 
>> >> >> From: Chris Heermann <heermann@EAST.ISI.EDU>
>> >> >> Subject: Re: 07/01/99 6Bone Routing Report
>> >> >> Date: Mon, 19 Jul 1999 10:27:11 -0400 (EDT)
>> >> >> Message-ID: <Pine.GSO.4.00.9907191000590.29470-100000@ale.east.isi.edu>
>> >> >>
>> >> >> 
>> >> >> Hi Masaki,
>> >> >> 
>> >> >> I hope I'm not being a pain, but I wanted to ping you about restarting
>> >> >> your BGP session with AS 1225 (cicnet). Can you please tell me how this
>> >> >> goes?
>> >> >> 
>> >> >> thanks,
>> >> >> Chris
>> >> >> 
>> >> >> On Sun, 4 Jul 1999, Masaki Hirabaru wrote:
>> >> >> 
>> >> >> > Hi. Chris,
>> >> >> > 
>> >> >> > I searched 1000::/3 in our dump data on July 1st. I couldn't find
>> >> >> > the beginning of announcements of 1000::/3. If you announced the
>> >> >> > route on that day, it should be recorded on this router with as
>> >> >> > path "237 7081".
>> >> >> > 
>> >> >> > Currently, we don't have 1000::/3 in our routing table, but AS
>> >> >> > 1225 (cicnet) says it comes through us (originated by CAIRN).
>> >> >> > According to MRTd's internal status, our router has never
>> >> >> > announced 1000::/3 to AS 109 (cisco) since the BGP session
>> >> >> > started 51 hours ago with cisco. 1000::/3 is flapping and I
>> >> >> > suspect something wrong along with the path.
>> >> >> > 
>> >> >> > Since July 1st, 1000::/3 has been announced as this and kept
>> >> >> > flapping. MRTd doesn't accept (but record) this announcement,
>> >> >> > because it detects a loop in the as path.
>> >> >> > 
>> >> >> > I could remove this by restarting a BGP session with cisco and/or
>> >> >> > cicnet, but I'll do that after I'm back from my vacation.
>> >> >> > 
>> >> >> > Masaki
>> >> >> > 
>> >> >> > [this is the first part of announcements of 1000::/3]
>> >> >> > BGP4+|07/01/99 12:04:49|A|3ffe:900:0:3::1|1225|1000::/3|1225 33 109 237 7081|IGP
>> >> >> > BGP4+|07/01/99 12:04:50|W|3ffe:900:0:3::1|1225|1000::/3
>> >> >> > BGP4+|07/01/99 12:06:24|A|3ffe:900:0:3::1|1225|1000::/3|1225 33 109 237 7081|IGP
>> >> >> > BGP4+|07/01/99 12:06:25|W|3ffe:900:0:3::1|1225|1000::/3
>> >> >> > BGP4+|07/01/99 12:17:27|A|3ffe:900:0:3::1|1225|1000::/3|1225 33 109 237 7081|IGP
>> >> >> > BGP4+|07/01/99 12:17:28|W|3ffe:900:0:3::1|1225|1000::/3
>> >> >> > BGP4+|07/01/99 12:26:30|A|3ffe:900:0:3::1|1225|1000::/3|1225 33 109 237 7081|IGP
>> >> >> > BGP4+|07/01/99 12:26:31|W|3ffe:900:0:3::1|1225|1000::/3
>> >> >> > BGP4+|07/01/99 12:27:09|W|3ffe:900:0:3::1|1225|1000::/3
>> >> >> > BGP4+|07/01/99 12:27:12|W|3ffe:900:0:3::1|1225|1000::/3
>> >> >> > BGP4+|07/01/99 12:40:45|A|3ffe:900:0:3::1|1225|1000::/3|1225 33 109 237 7081|IGP
>> >> >> > BGP4+|07/01/99 12:40:46|W|3ffe:900:0:3::1|1225|1000::/3
>> >> >> > [continues]
>> >> >> > 
>> >> >> > >> From: Craig Labovitz <labovit@HOME.MERIT.EDU>
>> >> >> > >> Subject: Re: 07/01/99 6Bone Routing Report
>> >> >> > >> Date: Fri, 02 Jul 1999 10:11:00 -0400
>> >> >> > >> Message-ID: <3.0.6.32.19990702101100.007d8240@HOME.MERIT.EDU>
>> >> >> > >>
>> >> >> > >> >Delivered-To: ipma-support@merit.edu
>> >> >> > >> >Date: Fri, 2 Jul 1999 09:24:21 -0400 (EDT)
>> >> >> > >> >From: Chris Heermann <heermann@EAST.ISI.EDU>
>> >> >> > >> >To: ipma-support@merit.edu
>> >> >> > >> >Cc: Chris Heermann <heermann@EAST.ISI.EDU>
>> >> >> > >> >Subject: Re: 07/01/99 6Bone Routing Report
>> >> >> > >> >
>> >> >> > >> >
>> >> >> > >> >Yesterday's report shows CAIRN advertising 0000::/0 and 1000::/3.
>> >> >> > >> >Apparently, we adertised to AS237, I think that's you/MERIT, and it came
>> >> >> > >> >back to you from AS1225. I have no clue as to how we advertised those two 
>> >> >> > >> >prefixes. Unless there was a momentary window when someone was
>> >> >> > >> >experimenting and got creative.
>> >> >> > >> >
>> >> >> > >> >Can you please provide me with the time this event occurred and any other
>> >> >> > >> >helpful details. thanks, Chris
>> >> >> > >> >
>> >> >> > >> >
>> >> >> > >> >> Non-6Bone Prefixes (outside of 3ffe::/16):
>> >> >> > >> >> --------------------------------
>> >> >> > >> >>   0000::/0 path 1225 33 109 237 7081 (CAIRN)  
>> >> >> > >> >>   1000::/3 path 1225 33 109 237 7081 (CAIRN)
>> >> >> > >> >
>> >> >> > >> >
>> >> >> > >> >
>> >> >> > >> >
>> >> >> > >> >
>> >> >> > >> 
>> >> >> > >> -------------------------
>> >> >> > >> Craig Labovitz					(734) 764-0252 voice
>> >> >> > >> Merit Network, Inc.				labovit@merit.edu
>> >> >> > >> 4251 Plymouth Road			
>> >> >> > >> Ann Arbor, MI 48105
>> >> >> >