We've been seeing some curious behaviour as a result of
equal-cost connected routes for interfaces participating
in HSRP being in the routing table. I have always been
under the impression that per-prefix load-sharing was
the default behaviour, unless doing process-switching,
or unless explicitly configured on a per-interface basis
under CEF switching.
What we see is this, which suggests that what is in fact
happening is per-packet load-sharing. Can anyone shed any
light on what might be going on here - could it be a bug
perhaps? If I try a TCP connection with packet debugging
turned on, per-prefix load-sharing does appear to happen
as expected. All the routers involved are running 12.0(8),
and there doesn't appear to be anything in the release
notes to suggest this is anything other than normal.
131.111.8.0/24 is a connected route on both cent-8 and
down-3, and cent-8, down-3, and west-3 are all attached
to a broadcast network, 131.111.2.0/24.
route-west-3.cam.ac.uk>trace 131.111.8.46
Type escape sequence to abort.
Tracing the route to www.cam.ac.uk (131.111.8.46)
1 route-down-3.cam.ac.uk (131.111.2.9) 4 msec
route-cent-8.cam.ac.uk (131.111.2.8) 0 msec
route-down-3.cam.ac.uk (131.111.2.9) 4 msec
2 aquila.csx.cam.ac.uk (131.111.8.74) 0 msec * 0 msec
M.
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:24 EDT