[ovs-dev] [PATCH v2] netdev-dpdk: Track vhost tx contention.

[Ilya Maximets]

On 20.10.2019 14:31, David Marchand wrote:
> Hello,
> 
> On Tue, Oct 15, 2019 at 2:13 PM Ilya Maximets <i.maximets at ovn.org> wrote:
>>
>> On 14.10.2019 20:04, Aaron Conole wrote:
>>> David Marchand <david.marchand at redhat.com> writes:
>>>
>>>> Add a coverage counter to help diagnose contention on the vhost txqs.
>>>> This is seen as dropped packets on the physical ports for rates that
>>>> are usually handled fine by OVS.
>>>> Document how to further debug this contention with perf.
>>>>
>>>> Signed-off-by: David Marchand <david.marchand at redhat.com>
>>>> ---
>>>> Changelog since v1:
>>>> - added documentation as a bonus: not sure this is the right place, or if it
>>>>     really makes sense to enter into such details. But I still find it useful.
>>>>     Comments?
>>>
>>> It's useful, and I think it makes sense here.
>>
>> That's an interesting debug method, but it looks not very suitable
>> for an end-user documentation.  One thing that bothers me the most
>> is referencing C code snippets in this kind of documentation.
> 
> Ok, can we conclude on the coverage counter wrt the v1 then?
> https://patchwork.ozlabs.org/patch/1153238/
> 
> Should I submit a v3 with the doc update (but removing the parts about perf) ?

'perf' part should not be there.

I'm in doubt about the rest of the docs.  This part might be useful, but
it doesn't provide any solution for the problem and I really don't know
if there is something we can suggest in that case.  "Rework your network
topology" doesn't sound like a friendly solution or exact steps to do.

One more thing is that documenting coverage counters doesn't look like a
good idea to me and I'd like to not create a precedent.

One day we'll rework this to be some "PMD/netdev performance statistics"
and it'll be OK to document it.  But there is nothing more permanent
than a temporary solution.

Right now the easiest way for me is to just apply v1.

Best regards, Ilya Maximets.