Re: [PATCH net-next v3] net: ipv4: add support for ECMP hash policy choice

From: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
To: nicolas.dichtel@6wind.com
Cc: David Ahern <dsa@cumulusnetworks.com>,
	Stephen Hemminger <stephen@networkplumber.org>,
	Roopa Prabhu <roopa@cumulusnetworks.com>,
	David Miller <davem@davemloft.net>,
	"netdev@vger.kernel.org" <netdev@vger.kernel.org>,
	jkbs@redhat.com, Eric Dumazet <edumazet@google.com>,
	Peter Christensen <pch@ordbogen.com>
Subject: Re: [PATCH net-next v3] net: ipv4: add support for ECMP hash policy choice
Date: Wed, 15 Mar 2017 12:46:23 +0200	[thread overview]
Message-ID: <B403ACFE-A028-454D-94C1-F7023457C87C@cumulusnetworks.com> (raw)
In-Reply-To: <5b5cf0ac-d152-3c8e-d54b-1059f561efee@6wind.com>

> On Mar 15, 2017, at 11:17 AM, Nicolas Dichtel <nicolas.dichtel@6wind.com> wrote:
> 
> Le 15/03/2017 à 00:45, David Ahern a écrit :
>> On 3/14/17 5:27 PM, Stephen Hemminger wrote:
>>> On Tue, 14 Mar 2017 15:38:40 -0700
>>> Roopa Prabhu <roopa@cumulusnetworks.com> wrote:
>>> 
>>>>>>>>> That's what my initial version did, but this was discussed during NetConf in Seville
>>>>>>>>> and it was decided that it's best to make a global sysctl, thus the change.  
>>>>>>>> 
>>>>>>>> Correct, we discussed this, and we all agreed to only have a sysctl for now.  
>>>>>>> 
>>>>>>> Why? If you are going to have private discussions please post the rationale
>>>>>>> in public.  
>>>>>> 
>>>>>> Stephen, is there any reason to have a per ecmp route multipath algo
>>>>>> selection ?.
>>>>>> All platforms have a global multipath selection algo. I also don't see
>>>>>> routing daemons ready or willing to specify a per ecmp route multipath
>>>>>> selection algo attribute.  
>>>>> 
>>>>> There is no compelling reason to make the attribute per route. But the
>>>>> issue is more that configuration through sysctl's is problematic. It doesn't
>>>>> fit into the standard API paradigm. Sysctl's are like routing patches not
>>>>> part of the real CLI. Trying to trap sysctl's for things like switchedev
>>>>> offload is particularly problematic. I can see the case for either way,
>>>>> and don't have a fixed opinion.  
>>>> 
>>>> ok. understand the switchdev offload part. It was that way in the past...but
>>>> today you can listen to sysctl updates on the netconf netlink channel.
>>>> it works pretty well.
>>> 
>>> Is there another patch to add the NETCONFA_ECMP support?
>>> 
>> 
>> does userspace care?
> Yes, I think it is needed so that userspace can correctly monitor this behavior.
> It also enables to check this parameter through netlink.
> 
> 
> Regards,
> Nicolas

This doesn’t fit the NETCONFA model well, there is no “all”, “default” or per iface option to be set, also
the other multipath sysctl which affects behaviour doesn’t have a notification.

Thanks,
 Nik