<html><head></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; "><div><br></div><div><br></div><div><span class="Apple-tab-span" style="white-space:pre"> </span>Creo que este mensaje de RIPE NCC es importante para la comunidad de ruteo. No se si a alguno le haya impactado en sus operaciones.</div><div><br></div><div>Slds</div><div>asn</div><div><br></div>Date: Fri, 27 Aug 2010 11:42:17 -0700 (PDT)<br>From: Lucy Lynch <<a href="mailto:llynch@civil-tongue.net">llynch@civil-tongue.net</a>><br>Subject: Re: Did your BGP crash today?<br>To: Grzegorz Janoszka <<a href="mailto:Grzegorz@Janoszka.pl">Grzegorz@Janoszka.pl</a>><br>Cc: <a href="mailto:nanog@nanog.org">nanog@nanog.org</a><br>Message-ID: <<a href="mailto:alpine.BSF.2.00.1008271141540.79214@hiroshima.bogus.com">alpine.BSF.2.00.1008271141540.79214@hiroshima.bogus.com</a>><br>Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed<br><br>FYI:<br><br>----------------------------------------------------------------------<br>Dear Colleagues,<br><br>On Friday 27 August, from 08:41 to 09:08 UTC, the RIPE NCC Routing<br>Information Service (RIS) announced a route with an experimental BGP<br>attribute. During this announcement, some Internet Service Providers<br>reported problems with their networking infrastructure.<br><br>Investigation<br>--------------<br><br>Immediately after discovering this, we stopped the announcement and<br>started investigating the problem. Our investigation has shown that the<br>problem was likely to have been caused by certain router types<br>incorrectly modifying the experimental attribute and then further<br>announcing the malformed route to their peers. The announcements sent<br>out by the RIS were correct and complied to all standards.<br><br>The experimental attribute was part of an experiment conducted in<br>collaboration with a group from Duke University. This involved<br>announcing a large (3000 bytes) optional transitive attribute, using a<br>modified version of Quagga. The attribute used type code 99. The data<br>consisted of zeros. We used the prefix 93.175.144.0/24 for this and<br>announced from AS 12654 on AMS-IX, NL-IX and GN-IX to all our peers.<br><br>Reports from affected ISPs showed that the length of the attribute in<br>the attribute header, as seen by their routers, was not correct. The<br>header stated 233 bytes and the actual data in their samples was 237<br>bytes. This caused some routers to drop the session with the peer that<br>announced the route.<br><br>We have built a test set-up which is running identical software and<br>configurations to the live set-up. From this set-up, and the BGP packet<br>dumps as made by the RIS, we have determined that the length of the data<br>in the attribute as sent out by the RIS was indeed 3000 bytes and that<br>all lengths recorded in the headers of the BGP updates were correct.<br><br>Beyond the RIS systems, we can only do limited diagnosis. One possible<br>explanation is that the affected routers did not correctly use the<br>extended length flag on the attribute. This flag is set when the length<br>of the attribute exceeds 255 bytes i.e. when two octets are needed to<br>store the length.<br><br>It may be that the routers may not add the higher octet of the length to<br>the total length, which would lead, in our test set-up, to a total<br>packet length of 236 bytes. If, in addition, the routers also<br>incorrectly trim the attribute length, the problem could occur as<br>observed. It is worth noting that the difference between the reported<br>233 and 237 bytes is the size of the flags, type code and length in the<br>attribute.<br><br>We will be further investigating this problem and will report any<br>findings. We regret any inconvenience caused.<br><br>Kind regards,<br><br>Erik Romijn<br><br>Information Services<br>RIPE NCC<br>_______________________________________________<br>tech-l mailing list<br><a href="mailto:tech-l@ams-ix.net">tech-l@ams-ix.net</a><br><a href="http://melix.ams-ix.net/mailman/listinfo/tech-l">http://melix.ams-ix.net/mailman/listinfo/tech-l</a><br><br><br><br>- Lucy</body></html>