On 3/16/23 21:00, Ilya Maximets wrote:
> On 3/16/23 20:44, Dumitru Ceara wrote:
>>> diff --git a/ovn-nb.xml b/ovn-nb.xml
>>> index 73f707aa0..c5dbebd1d 100644
>>> --- a/ovn-nb.xml
>>> +++ b/ovn-nb.xml
>>> @@ -2041,6 +2041,13 @@ or
>>> the affinity timeslot. Max supported affinity_timeout is 65535
>>> seconds.
>>> </column>
>>> +
>>> + <column name="options" key="ct_flush_enabled"
>>> + type='{"type": "boolean"}'>
>>
>> As Ilya mentioned, _enabled is probably redundant.
>>
>> I wonder though whether we need a different name. I'm not sure if it's
>> better to be honest but what do you think of "graceful_cleanup".In the
>> end what we achieve with ct_flush_enabled=false is "graceful
>> termination" of the connection.
>
> This is questionable. I'd argue that flush is a cleaner option, so
> it is more graceful. :)
>
> Jokes aside, OVN doesn't know what CMS/user is going to do with the
> backend/pod/VM. Form pure OVN perspective we're just removing a
> beckend/LB. I would not expect an option named '*_cleanup=false' to
> perform some extra actions, as well as '*_cleanup=true' to do nothing.
> Unless, of course, it is a 'do_not_cleanup' option.
>
Fair points above. I'd say let's keep "ct_flush=true/false" then.
> Best regards, Ilya Maximets.
>
Thanks,
Dumitru
_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
