[sfc-dev] Service Chain Last Hop

João Silva silvajp9 at gmail.com
Wed Aug 26 16:19:08 UTC 2015


That last rule 'throws' back the packet received from the last service
function to the port where it was received from (IN_PORT). The packet is
sent as received (including VXLANGPE + NSH encap) except for the nshc1
which is set to 0.

Now when the last sf receives this 're-sent' packet it just ignores it
because the mac_dst is the SFF, and the packet dies here never reaching its
final destination.

Maybe the nshc1 = 0 is a flag for the python sff agent to decap and forward
the packet. But on the OVS I would expect the last rule to either:

   - pop the headers and forward
   - or forward to the classifier

To comply with what is being discussed here as the correct behavior for the
last hop.

Is this rule a bug, the last hop for the ovs sff is not yet implemented or
a config problem?
Can someone shed some light on this?

Best regards,

João
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opendaylight.org/pipermail/sfc-dev/attachments/20150826/64b93035/attachment.html>


More information about the sfc-dev mailing list