On 3/1/23 23:42, Ilya Maximets wrote:
> On 2/28/23 16:28, Dumitru Ceara wrote:
>> On 2/28/23 16:22, Ilya Maximets wrote:
>>> On 2/28/23 16:14, Dumitru Ceara wrote:
On 2/28/23 15:29, Ilya Maximets wrote:
> On 2/28/23 01:03, Dumitru Ceara wrote:
>> Hi Ilya,
>>
>> Thanks for the p
On 2/28/23 16:28, Dumitru Ceara wrote:
> On 2/28/23 16:22, Ilya Maximets wrote:
>> On 2/28/23 16:14, Dumitru Ceara wrote:
>>> On 2/28/23 15:29, Ilya Maximets wrote:
On 2/28/23 01:03, Dumitru Ceara wrote:
> Hi Ilya,
>
> Thanks for the patch, you're right the hairpin SNAT IP processi
On 2/28/23 16:22, Ilya Maximets wrote:
> On 2/28/23 16:14, Dumitru Ceara wrote:
>> On 2/28/23 15:29, Ilya Maximets wrote:
>>> On 2/28/23 01:03, Dumitru Ceara wrote:
Hi Ilya,
Thanks for the patch, you're right the hairpin SNAT IP processing in
ovn-controller is extremely ineffici
On 2/28/23 16:14, Dumitru Ceara wrote:
> On 2/28/23 15:29, Ilya Maximets wrote:
>> On 2/28/23 01:03, Dumitru Ceara wrote:
>>> Hi Ilya,
>>>
>>> Thanks for the patch, you're right the hairpin SNAT IP processing in
>>> ovn-controller is extremely inefficient today.
>>>
>>> I didn't review the code clo
On 2/28/23 15:29, Ilya Maximets wrote:
> On 2/28/23 01:03, Dumitru Ceara wrote:
>> Hi Ilya,
>>
>> Thanks for the patch, you're right the hairpin SNAT IP processing in
>> ovn-controller is extremely inefficient today.
>>
>> I didn't review the code closely but I do have some initial remarks below.
>
On 2/28/23 01:03, Dumitru Ceara wrote:
> Hi Ilya,
>
> Thanks for the patch, you're right the hairpin SNAT IP processing in
> ovn-controller is extremely inefficient today.
>
> I didn't review the code closely but I do have some initial remarks below.
>
> On 2/20/23 12:42, Ilya Maximets wrote:
>>
Hi Ilya,
Thanks for the patch, you're right the hairpin SNAT IP processing in
ovn-controller is extremely inefficient today.
I didn't review the code closely but I do have some initial remarks below.
On 2/20/23 12:42, Ilya Maximets wrote:
> It's common to have 'hairpin_snat_ip' to be configured
It's common to have 'hairpin_snat_ip' to be configured exactly the
same for each and every Load Balancer in a setup. For example,
ovn-kubernetes does that, setting '169.254.169.5 fd69::5' value
unconditionally for every LB:
https://github.com/ovn-org/ovn-kubernetes/blob/cd78ae1af4657d38bdc41003
