On 08/20/2015 4:00 PM, David Farmer wrote:
Sorry guys that was us, or one of our participants.
They made some changes about an hour ago, and announced the whole IPv6 table to us. We had a prefix list configured, but it seems to be empty now, not sure how or when that happened.
Ah, that explains it. I found the other missing piece in our configuration and Rob was right about filters. We've got an explicit deny against ^6939_ from the route server because of our bilateral peering with HE and that's keeping our IPv6 routes from the route server down to 45 and 49 routes respectively so setting a limit of 150 is within reason. But the routes you sent wouldn't have matched the filter, hence the prefix limit was exceeded. Everyone else must have a 13K+ limit and therefor didn't notice the additional paths being injected. Or folks haven't put a prefix-limit on their IPv6 neighbors because if this was a full table, that's nearly 23K routes... -James
We're upgrading to ASRs soon, and they have a check for empty prefix lists.
On 8/20/15 15:12 , James Stahr wrote:
Is someone just start re-advertising HE via the route server or did we just configure today a session for HE to the route server? It was my impression that they only did bilateral peering.
Aug 20 14:04:01.902 CDT: %BGP-4-MAXPFX: Number of prefixes received from 2001:504:27::D1AF:0:1 (afi 1) reaches 143, max 150 Aug 20 14:04:01.908 CDT: %BGP-3-MAXPFXEXCEED: Number of prefixes received from 2001:504:27::D1AF:0:1 (afi 1): 151 exceeds limit 150
Our session is currently down and will remain down. When the session was up, we saw all of the paths start with 6939.
James Stahr CDW