23 Mar
2020
23 Mar
'20
3:24 p.m.
Joe Maimon wrote:
Seems to be finding plenty of these
AGGREGATOR attribute is BGP_AS_ZERO(0)
However, rfc7607 specifies error handling as per rfc7606 which suggests most frequently the withdrawn treatment
bgpd appears to instead proceed for ebgp peers and notify for ibgp sessions, I am curious as to how either of these are correct and would appreciate any advice on the matter.
Frustratingly, I haven't figured out any good way to log which prefix update are triggering this.
Best,
Joe
This seems to be working out for now. Just a thought, but if frr behavior for error handling isnt strictly rfc driven, perhaps it should be configurable, sort of relaxed mode or strict mode, by peer even.