]> git.puffer.fish Git - mirror/frr.git/commit
IPv6 multipath is broken in BGP if nexthop contains only global address.
authorDonald Sharp <sharpd@cumulusnetworks.com>
Wed, 20 May 2015 00:24:45 +0000 (17:24 -0700)
committerDonald Sharp <sharpd@cumulusnetworks.com>
Wed, 20 May 2015 00:24:45 +0000 (17:24 -0700)
commit000830bdaec70daeeefc1289c1da22e9b7e7c9dc
tree10d5d48379eb0106fa56fc798d8b84fde4fc726d
parent431aa9f906144968a9ca6c127da292f46ffa242e
IPv6 multipath is broken in BGP if nexthop contains only global address.
IPv6 always uses both nextop IPv6 address and ifIndex in sending routes down to
zebra. In cases where only the global IPv6 address is present in the nexthop
information, the existing code doesn't set the ifIndex. An example of such a
case is when a route-map isused with "set ipv6 next-hop" and only global
address is specified. This code causes the ifIndex to be determined and
set thereby fixing the multipath programming.

Signed-off-by: Dinesh G Dutt <ddutt@cumulusnetworks.com>
Reviewed-by: Shrijeet Mukherjee <shm@cumulusnetworks.com>
bgpd/bgp_zebra.c