Native userspace tunnels learn tnl-neigh from the outer Ethernet
source MAC of every encapsulated IPv4/IPv6 data packet. That MAC is
the last hop, not the remote VTEP. Under ECMP or MLAG the cache
flaps, seq_change(tnl_conf_seq) runs, and every datapath flow is
revalidated.
Add Interface option neigh_snoop, default true, so BFD still recovers
when the remote VTEP NIC MAC actually changes. Setting
options:neigh_snoop=false on an OpenFlow tunnel port skips the
data-packet tnl_neigh_set() path for that port only. ARP, GARP, and
ND snooping is unchanged.
Userspace native tunneling only. Tests: tunnel_push_pop,
tunnel_push_pop_ipv6.
v2: Resend via git send-email. v1 failed git am (Gmail linkified
tests/tunnel-push-pop*.at paths).
Girish Kumar (1):
xlate: Add neigh_snoop option to skip ECMP MAC flaps.
NEWS | 6 ++
lib/netdev-vport.c | 10 +++
lib/netdev.h | 1 +
lib/tnl-ports.c | 27 +++++++-
lib/tnl-ports.h | 1 +
ofproto/ofproto-dpif-xlate.c | 116 +++++++++++++++++++++++++++++++++-
ofproto/tunnel.c | 36 +++++++++++
ofproto/tunnel.h | 3 +
tests/tunnel-push-pop-ipv6.at | 48 ++++++++++++++
tests/tunnel-push-pop.at | 48 ++++++++++++++
vswitchd/vswitch.xml | 13 ++++
11 files changed, 306 insertions(+), 3 deletions(-)
Signed-off-by: Girish Kumar <[email protected]>
--
2.47.3
_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
