Re: [PATCH] multifd: Set a higher "backlog" default value for listen()

[Juan Quintela]

"Wang, Wei W" <wei.w.wang@intel.com> wrote:
> On Friday, May 19, 2023 10:52 AM, Wang, Lei4 wrote:
>> > We can change it to uint16_t or uint32_t, but need to see if listening
>> > on a larger value is OK to everyone.
>> 
>> Is there any use case to use >256 migration channels? If not, then I suppose
>> it's no need to increase it.
>
> People can choose to use more than 256 channels to boost performance.

See my other email, I doubt it any time soon O:-)

> If it is determined that using larger than 256 channels doesn't increase 
> performance
> on all the existing platforms, then we need to have it reflected in the code 
> explicitly,
> e.g. fail with errors messages when user does:
> migrate_set_parameter multifd-channels 512


(qemu) migrate_set_parameter multifd-channels 300
Error: Parameter 'multifd-channels' expects uint8_t

So I think that is working.

>> 
>> >
>> > Man page of listen mentions that the  maximum length of the queue for
>> > incomplete sockets can be set using
>> > /proc/sys/net/ipv4/tcp_max_syn_backlog,
>> > and it is 4096 by default on my machine