We would like to allow using LRP IPs as a source IP for the LB Health Checks.
The main motivation is that with the current implementation, the CMS has to set
aside at least one IP per LSP to be later used for health check probes and
that's in the very least impractical. It also complicates introduction of the
health check feature in existing deployments that were not designed with this
limitation in mind.

Overall we believe that if this proposal is viable, and LRP IPs could be used
for health checks, it would simplify management of health checks from the user
perspective.

So far I only did manual testing and it doesn't seem to break anything, so I was
interested in the community's opinion on this approach. The only scenario that
comes to mind that would be "broken" is if the LB backend was sitting on
the public side of SNAT and the "real" traffic to backend would be using LRP's
IP. However that seems like an unlikely setup, and it's a smaller constraint
than the current state.

Thank you for a feedback,
Martin.

Martin Kalcok (1):
  northd: Allow LB health checks from router IPs.

 northd/northd.c | 95 +++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 93 insertions(+), 2 deletions(-)

-- 
2.51.0

_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev

Reply via email to