Hi Lorenzo,
Thanks for addressing my concerns so quickly. I see that you addressed
my concerns about load balancers with no ports in patch 3. However there
are still some lingering issues:
* In patch 1, the commit_lb_aff() action still expects all parsed IP
addresses to have a port.
* In patch 1, the commit_lb_aff() action always expects a protocol to be
present when parsing.
* Because of the above 2 issues, the resulting learn() action that we
create when encoding the commit_lb_aff() action will always specify a
port number and protocol.
* In patch 3, northd always sets the protocol in the commit_lb_aff()
action it creates. It should only specify a protocol if a port is
specified on either the LB VIP or backend.
On 10/13/22 11:51, Lorenzo Bianconi wrote:
Introduce load-balancer affinity timeout in order to dnat connections received
from the same client to a given load-balancer to the same backend if received
in the affinity timeslot specified by the CMS.
Reported-at: https://bugzilla.redhat.com/show_bug.cgi?id=2127959
Changes since v1:
- rebase on top of ovn master
- fix system-ovn.at
- fix documentation
- take into account the case the CMS does not provide vip/backends ports
Lorenzo Bianconi (3):
actions: introduce commit_lb_aff action
actions: introduce chk_lb_aff action
northd: rely on new actions for lb affinity
controller/lflow.h | 1 +
include/ovn/actions.h | 16 ++
include/ovn/logical-fields.h | 3 +
lib/actions.c | 389 +++++++++++++++++++++++++++++++++++
lib/lb.c | 3 +
lib/lb.h | 1 +
northd/northd.c | 202 +++++++++++++++---
northd/ovn-northd.8.xml | 208 ++++++++++++++++---
ovn-nb.xml | 6 +
ovn-sb.xml | 44 ++++
tests/ovn-northd.at | 346 +++++++++++++++++++------------
tests/ovn.at | 65 +++---
tests/system-ovn.at | 184 ++++++++++++++++-
utilities/ovn-trace.c | 4 +
14 files changed, 1251 insertions(+), 221 deletions(-)
_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
