Hi , On a MPLS network offering L3 VPN services , There is a situation where a MP-eBGP update with more than 128 extended communities RTs brings down VPNv4 sessions between the RR and 7600 PEs due to a bug CSCee30718 on old code of 7600.
On the old code if an MP-BGP update with more than 128 RTs or 255 standard communities will reset the session . On the new codes the limit has been lifted. Now the only solution to prevent these situation is upgrade to a newer code or do filtering for communities at edge where I have some findings & questions. - Extended communities RTs also can be set on a IPv4 unicast session by a CE towards PE of a MPLS network and RTs get exported into VRF & network automatically . PE has to filter these. Not sure how many SPs do this ? - On a session a BGP update for a prefix can have lots of standard and extended communities , Is there a limit ? - If an unauthorized BGP speaker sends prefixes with lots of standard and extended communities on it can eat it up memory ? I see a need for a CLI knob to drop updates with once the number of communities crosses a limit on a update like one already for ASPATH - bgp maxas-limit. Any suggestions or ideas ? Thanks !! Dan _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
