Hi Eli and Tim Eli Britstein <[email protected]> writes:
> CT userspace behavior is not aligned with the kernel's one in some > cases. Handle it. > > Note: this series takes over [1] as related to the same topic. > > [1] > https://patchwork.ozlabs.org/project/openvswitch/patch/[email protected]/ As a general comment on the series, there are some needed NEWS entries for the work here. It is quite user visible behavior change and really needs to be documented as such. As mentioned in other thread, these are not Fixes - they are functional changes. I wouldn't generally consider these as something to backport to older branches unless there were many requests. I did take a look at some of the OSP Neutron code, but it does seem like most of neutron has moved on to OVN. Regardless, there may be deployments that we're not aware and have their flows relying on the older CT behavior, so we need to be very explicit when documenting these changes. > Eli Britstein (2): > conntrack: Align userspace NAT binding on ct(commit) with kernel. > conntrack: Align hairpin reply lookup and NAT with kernel. > > Tim Rozet (1): > conntrack: Match kernel NAT handling for new packets. > > lib/conntrack.c | 287 +++++++++++++++++++++++++++---- > tests/system-kmod-macros.at | 45 +++++ > tests/system-traffic.at | 191 ++++++++++++++++++++ > tests/system-userspace-macros.at | 9 + > tests/test-conntrack.c | 117 +++++++++++++ > 5 files changed, 618 insertions(+), 31 deletions(-) _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
