[c-nsp] BFD CPU hog and traceback on me3600 and isis flapping 15.3(3)S1a

[Pete Lumbis]

Doesn't matter. The trigger is the FRR event that triggers a walk of the
forwarding table, even if there isn't a change. The process of walking the
table is done under the BFD process (incorrectly) and causes the hog and
could starve out the Hello thread.


On Wed, Mar 19, 2014 at 9:06 AM, Mark Tinka <mark.tinka at seacom.mu> wrote:

> On Wednesday, March 19, 2014 02:56:02 PM Pete Lumbis wrote:
>
> > Just to close the loop on the thread, I spoke to OP
> > off-list and this matches CSCug77067. When an FRR event
> > (either TE FRR or IP FRR) occurs the processes
> > prioritization for BFD and the FRR event doesn't do what
> > it's supposed to and BFD triggers a CPU hog. It's also
> > possible, as Adam saw, for BFD neighbors to flap. This
> > is fixed in 15.4.1S1 and 15.3.3S2
>
> Pete, is this only when BFD is enabled for the IGP, or when
> BFD is enabled for RSVP-TE?
>
> Or does this not matter at all?
>
> Mark.
>