[ovs-dev] [PATCH ovn v2] Revert features detection and zero-snat patches.

Dumitru Ceara dceara at redhat.com
Mon Aug 30 07:38:23 UTC 2021


On 8/28/21 2:44 AM, Numan Siddique wrote:
> On Fri, Aug 27, 2021 at 8:06 PM Mark Michelson <mmichels at redhat.com> wrote:
>>
>> Thank you both for the quick reviews. I pushed the change to master.

Thanks Mark, Numan, and Flavio for fixing this and sorry for introducing
the issues in the first place.

> 
> Would this require a backport to branch-21.06 ?
> 

If it's not too much work this backport might be a good to have, to
avoid behavioral differences between released versions.

I'll start thinking about a different way of addressing what 58683a42
(ovn-controller: Handle DNAT/no-NAT conntrack tuple collisions.) was
trying to fix.

> Numan
> 

Regards,
Dumitru

>>
>> On 8/27/21 5:28 PM, Flavio Fernandes wrote:
>>>
>>>
>>>> On Aug 27, 2021, at 4:11 PM, Mark Michelson <mmichels at redhat.com
>>>> <mailto:mmichels at redhat.com>> wrote:
>>>>
>>>> This commit reverts the following two commits:
>>>> 56e2cd3a (ovn-controller: Detect OVS datapath capabilities.)
>>>> 58683a42 (ovn-controller: Handle DNAT/no-NAT conntrack tuple
>>>> collisions.)
>>>>
>>>> The former commit causes an incompatibility with OVS databases that do
>>>> not have the datapath table present. This specifically is making
>>>> OpenStack upgrades from OSP 13 to OSP 16.2 fail.
>>>>
>>>> The latter commit causes significantly degraded dataplane performance
>>>> when testing a sense OpenShift cluster. This was pinpointed to be due to
>>>> an extra ct(nat(src)) that this commit added.
>>>>
>>>> Reported-at: https://bugzilla.redhat.com/show_bug.cgi?id=1992705
>>>> <https://bugzilla.redhat.com/show_bug.cgi?id=1992705>
>>>> Reported-at: https://bugzilla.redhat.com/show_bug.cgi?id=1992012
>>>> <https://bugzilla.redhat.com/show_bug.cgi?id=1992012>
>>>>
>>>> Signed-off-by: Mark Michelson <mmichels at redhat.com
>>>> <mailto:mmichels at redhat.com>>
>>>>
>>>> ---
>>>> v1 -> v2:
>>>>  * Switched from reverting just one commit to reverting two, thereby
>>>>    eliminating two issues.
>>>> ---
>>>>
>>>> Signed-off-by: Mark Michelson <mmichels at redhat.com
>>>> <mailto:mmichels at redhat.com>>
>>>> ---
>>>
>>>
>>> Acked-by: Flavio Fernandes <flavio at flaviof.com <mailto:flavio at flaviof.com>>
>>>
>>>
>>>> controller/ovn-controller.c   | 115 +++++---------------
>>>> include/ovn/actions.h         |   1 -
>>>> include/ovn/features.h        |  18 ----
>>>> lib/actions.c                 |  31 ------
>>>> lib/automake.mk               |   1 -
>>>> lib/features.c                |  84 ---------------
>>>> lib/test-ovn-features.c       |  56 ----------
>>>> tests/automake.mk             |   3 -
>>>> tests/ovn-controller.at <http://ovn-controller.at>       |  11 +-
>>>> tests/ovn-features.at <http://ovn-features.at>         |   8 --
>>>> tests/ovn.at <http://ovn.at>                  |   2 +-
>>>> tests/system-common-macros.at <http://system-common-macros.at> |   4 -
>>>> tests/system-ovn.at <http://system-ovn.at>           | 190
>>>> ----------------------------------
>>>> tests/testsuite.at <http://testsuite.at>            |   1 -
>>>> 14 files changed, 34 insertions(+), 491 deletions(-)
>>>> delete mode 100644 lib/features.c
>>>> delete mode 100644 lib/test-ovn-features.c
>>>> delete mode 100644 tests/ovn-features.at <http://ovn-features.at>
>>>
>>
>> _______________________________________________
>> dev mailing list
>> dev at openvswitch.org
>> https://mail.openvswitch.org/mailman/listinfo/ovs-dev
>>
> _______________________________________________
> dev mailing list
> dev at openvswitch.org
> https://mail.openvswitch.org/mailman/listinfo/ovs-dev
> 



More information about the dev mailing list