[c-nsp] 6500 - stateful failover, reason?

[Brandon Ewing]

On Tue, Sep 22, 2009 at 01:07:47PM -0700, Bill Blackford wrote:
> <snip>
> This is normal -- the new processor has to rebuild all routing protocols when it comes online from RPR or RPR+ mode.  Even with SSO mode, if NSF is not configured with all peers on all protocols, the sessions are broken down and rebuilt.
> </snip>
> 
> I recently had event very much like this. In my case, even SSO/NSF dropped some adjacencies during this switchover.
> 
> -b
> 

Did you confirm that NSF (Graceful restart, etc) has been negotiated with
all adjacencies, and that all adjacent routers are NSF-aware?  

Also, there is an upper bound to the time alloted to
perform the switchover, and the new RP has to signal an NSF event prior to
current hold/dead/etc timers expiring -- if you have your OSPF timer set to
1/3, and it takes more than 3 seconds for the standby RP to see the failure,
take over, and send the NSF message, the adjacency will already have been
dropped.

-- 
Brandon Ewing                                        (nicotine at warningg.com)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <https://puck.nether.net/pipermail/cisco-nsp/attachments/20090922/036ae784/attachment.bin>