Re: send/receive data via eth0/1 external loop ?

Re: send/receive data via eth0/1 external loop ?

[Roland Kletzing]

great, thanks!


-----Ursprüngliche Nachricht-----
Von: "Julian Anastasov" <ja@ssi.bg>
Gesendet: 12.11.2011 19:31:13
An: "Roland Kletzing" <devzero@web.de>
Betreff: Re: send/receive data via eth0/1 external loop ?

>
> Hello,
>
>On Sat, 12 Nov 2011, Roland Kletzing wrote:
>
>> Hello,
>> i`m searching for a while now, but i did not find a solution, so forgive when
>> i`m (as a user) asking for expert help here.
>>
>> I want to do some network stress/burn-in and quality/reliability testing,
>> i.e. i want to send/receive large amounts of data trough the systems network
>> interfaces without the need of hogging the net or other systems.
>>
>> so, that test should run "loopback", i.e. connecting eth0+1 via crossover cable
>> or via locally attached switch. I don`t want to use a second system for that.
>>
>> testing at ip/tcp level would be favourable , but raw eth level would be also fine.
>>
>> But how ?
>> What appears simple on a first view, seems to be difficult.... :(
>>
>> All i came across is this report:
>> http://www.zyztematik.com/?p=10
>> and this description of a patch:
>> http://www.ssi.bg/~ja/send-to-self.txt
>>
>> Does somebody know a proper way of achieving this without patching the
>> kernel ?
>
> You can use accept_local flag and rule with pref 0:
>
>http://marc.info/?l=linux-netdev&m=125983955627148&w=2
>
>Regards
>
>--
>Julian Anastasov <ja@ssi.bg>


___________________________________________________________
SMS schreiben mit WEB.DE FreeMail - einfach, schnell und
kostenguenstig. Jetzt gleich testen! http://f.web.de/?mc=021192

Re: send/receive data via eth0/1 external loop ?

[Roland Kletzing]

to confirm: IT WORKS! (tested with 2.6.38 kernel)  :)

more detail:
>> You can use accept_local flag and rule with pref 0:
>>
>>http://marc.info/?l=linux-netdev&m=125983955627148&w=2

this in itself is not enought to get it running, but with this one
http://marc.info/?l=linux-netdev&m=129500124207256&w=2
(adapted for eth0/1), it looks good  and i see packets flow between eth0/1

thanks again for the pointer !

regards
Roland


-----Ursprüngliche Nachricht-----
Von: "Roland Kletzing" <devzero@web.de>
Gesendet: 12.11.2011 19:50:33
An: "Julian Anastasov" <ja@ssi.bg>
Betreff: Re: send/receive data via eth0/1 external loop ?

>great, thanks!
>
>
>-----Ursprüngliche Nachricht-----
>Von: "Julian Anastasov" <ja@ssi.bg>
>Gesendet: 12.11.2011 19:31:13
>An: "Roland Kletzing" <devzero@web.de>
>Betreff: Re: send/receive data via eth0/1 external loop ?
>
>>
>> Hello,
>>
>>On Sat, 12 Nov 2011, Roland Kletzing wrote:
>>
>>> Hello,
>>> i`m searching for a while now, but i did not find a solution, so forgive when
>>> i`m (as a user) asking for expert help here.
>>>
>>> I want to do some network stress/burn-in and quality/reliability testing,
>>> i.e. i want to send/receive large amounts of data trough the systems network
>>> interfaces without the need of hogging the net or other systems.
>>>
>>> so, that test should run "loopback", i.e. connecting eth0+1 via crossover cable
>>> or via locally attached switch. I don`t want to use a second system for that.
>>>
>>> testing at ip/tcp level would be favourable , but raw eth level would be also fine.
>>>
>>> But how ?
>>> What appears simple on a first view, seems to be difficult.... :(
>>>
>>> All i came across is this report:
>>> http://www.zyztematik.com/?p=10
>>> and this description of a patch:
>>> http://www.ssi.bg/~ja/send-to-self.txt
>>>
>>> Does somebody know a proper way of achieving this without patching the
>>> kernel ?
>>
>> You can use accept_local flag and rule with pref 0:
>>
>>http://marc.info/?l=linux-netdev&m=125983955627148&w=2
>>
>>Regards
>>
>>--
>>Julian Anastasov <ja@ssi.bg>
>


___________________________________________________________
SMS schreiben mit WEB.DE FreeMail - einfach, schnell und
kostenguenstig. Jetzt gleich testen! http://f.web.de/?mc=021192

send/receive data via eth0/1 external loop ?

[Roland Kletzing]

Hello,
i`m searching for a while now, but i did not find a solution, so forgive when
i`m (as a user) asking for expert help here.

I want to do some network stress/burn-in and quality/reliability testing,
i.e. i want to send/receive large amounts of data trough the systems network
interfaces without the need of hogging the net or other systems.

so, that test should run "loopback", i.e. connecting eth0+1 via crossover cable
or via locally attached switch. I don`t want to use a second system for that.

testing at ip/tcp level would be favourable , but raw eth level would be also fine.

But how ?
What appears simple on a first view, seems to be difficult.... :(

All i came across is this report:
http://www.zyztematik.com/?p=10
and this description of a patch:
http://www.ssi.bg/~ja/send-to-self.txt

Does somebody know a proper way of achieving this without patching the
kernel ?

regards
Roland
___________________________________________________________
SMS schreiben mit WEB.DE FreeMail - einfach, schnell und
kostenguenstig. Jetzt gleich testen! http://f.web.de/?mc=021192

Re: send/receive data via eth0/1 external loop ?

[Julian Anastasov]

	Hello,

On Sat, 12 Nov 2011, Roland Kletzing wrote:

> Hello,
> i`m searching for a while now, but i did not find a solution, so forgive when
> i`m (as a user) asking for expert help here.
> 
> I want to do some network stress/burn-in and quality/reliability testing,
> i.e. i want to send/receive large amounts of data trough the systems network
> interfaces without the need of hogging the net or other systems.
> 
> so, that test should run "loopback", i.e. connecting eth0+1 via crossover cable
> or via locally attached switch. I don`t want to use a second system for that.
> 
> testing at ip/tcp level would be favourable , but raw eth level would be also fine.
> 
> But how ?
> What appears simple on a first view, seems to be difficult.... :(
> 
> All i came across is this report:
> http://www.zyztematik.com/?p=10
> and this description of a patch:
> http://www.ssi.bg/~ja/send-to-self.txt
> 
> Does somebody know a proper way of achieving this without patching the
> kernel ?

	You can use accept_local flag and rule with pref 0:

http://marc.info/?l=linux-netdev&m=125983955627148&w=2

Regards

--
Julian Anastasov <ja@ssi.bg>