* NFQUEUE in bridge mode performance poor in the last kernels
@ 2012-03-06 2:05 tingwei liu
2012-03-06 15:19 ` Eric Dumazet
2012-03-06 21:23 ` Florian Westphal
0 siblings, 2 replies; 6+ messages in thread
From: tingwei liu @ 2012-03-06 2:05 UTC (permalink / raw)
To: netfilter-devel
In the past few days, I have puzzled by NFQUEUE in bridge mode.
I have take some test with five kernels.
2.6.24.4
2.6.36.4
2.6.38
3.0.8
3.1.10
The result is : 2.6.24.4,2.6.26.4,2.6.38 have a goog performance;
3.0.8 and 3.1.10 have a poor performance.
Next is copy from suricata maillist( eric@regit.org )
I'm having a look at it. There has been some changes between the two
kernel versions (bringing more performances) but it seems there is some
side effects with bridge.
Thanks!
--
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: NFQUEUE in bridge mode performance poor in the last kernels
2012-03-06 2:05 NFQUEUE in bridge mode performance poor in the last kernels tingwei liu
@ 2012-03-06 15:19 ` Eric Dumazet
2012-03-07 0:46 ` tingwei liu
2012-03-06 21:23 ` Florian Westphal
1 sibling, 1 reply; 6+ messages in thread
From: Eric Dumazet @ 2012-03-06 15:19 UTC (permalink / raw)
To: tingwei liu; +Cc: netfilter-devel
Le mardi 06 mars 2012 à 10:05 +0800, tingwei liu a écrit :
> In the past few days, I have puzzled by NFQUEUE in bridge mode.
> I have take some test with five kernels.
> 2.6.24.4
> 2.6.36.4
> 2.6.38
> 3.0.8
> 3.1.10
>
> The result is : 2.6.24.4,2.6.26.4,2.6.38 have a goog performance;
> 3.0.8 and 3.1.10 have a poor performance.
> Next is copy from suricata maillist( eric@regit.org )
>
> I'm having a look at it. There has been some changes between the two
> kernel versions (bringing more performances) but it seems there is some
> side effects with bridge.
> Thanks!
More data is welcomed. What is "good", what is "bad" ?
You dont expect us to magically understand the issue, do you ?
If you want some help, you should give as much information as possible.
--
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: NFQUEUE in bridge mode performance poor in the last kernels
2012-03-06 2:05 NFQUEUE in bridge mode performance poor in the last kernels tingwei liu
2012-03-06 15:19 ` Eric Dumazet
@ 2012-03-06 21:23 ` Florian Westphal
2012-03-07 1:26 ` tingwei liu
1 sibling, 1 reply; 6+ messages in thread
From: Florian Westphal @ 2012-03-06 21:23 UTC (permalink / raw)
To: tingwei liu; +Cc: netfilter-devel
tingwei liu <tingw.liu@gmail.com> wrote:
> In the past few days, I have puzzled by NFQUEUE in bridge mode.
> I have take some test with five kernels.
> 2.6.24.4
> 2.6.36.4
> 2.6.38
> 3.0.8
> 3.1.10
>
> The result is : 2.6.24.4,2.6.26.4,2.6.38 have a goog performance;
> 3.0.8 and 3.1.10 have a poor performance.
> Next is copy from suricata maillist( eric@regit.org )
>
> I'm having a look at it. There has been some changes between the two
> kernel versions (bringing more performances) but it seems there is some
> side effects with bridge.
Might be the 'gro+nfqueue eats MAC header' problem, you could try
commit a8db7b2d197a0d624baab83f0c810b0edbc4ffd0 (netfilter: nf_queue: fix
queueing of bridged gro skbs). Or, disable gro on all bridge ports via
ethtool -K $device gro off
If its not gro related, please provide more information about your
machine, setup, ... etc.
--
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: NFQUEUE in bridge mode performance poor in the last kernels
2012-03-06 15:19 ` Eric Dumazet
@ 2012-03-07 0:46 ` tingwei liu
2012-03-07 1:19 ` Eric Dumazet
0 siblings, 1 reply; 6+ messages in thread
From: tingwei liu @ 2012-03-07 0:46 UTC (permalink / raw)
To: Eric Dumazet; +Cc: netfilter-devel
On Tue, Mar 6, 2012 at 11:19 PM, Eric Dumazet <eric.dumazet@gmail.com> wrote:
> Le mardi 06 mars 2012 à 10:05 +0800, tingwei liu a écrit :
>> In the past few days, I have puzzled by NFQUEUE in bridge mode.
>> I have take some test with five kernels.
>> 2.6.24.4
>> 2.6.36.4
>> 2.6.38
>> 3.0.8
>> 3.1.10
>>
>> The result is : 2.6.24.4,2.6.26.4,2.6.38 have a goog performance;
>> 3.0.8 and 3.1.10 have a poor performance.
>> Next is copy from suricata maillist( eric@regit.org )
>>
>> I'm having a look at it. There has been some changes between the two
>> kernel versions (bringing more performances) but it seems there is some
>> side effects with bridge.
>> Thanks!
>
> More data is welcomed. What is "good", what is "bad" ?
>
Sorry for that.
My network is 100Mbps.
"good" means the forward bandwidth almost 100Mbps.
"bad" means the forward bandwidth with nfqueue only 30Mbps.
Thanks for your reply!
> You dont expect us to magically understand the issue, do you ?
>
> If you want some help, you should give as much information as possible.
>
>
>
--
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: NFQUEUE in bridge mode performance poor in the last kernels
2012-03-07 0:46 ` tingwei liu
@ 2012-03-07 1:19 ` Eric Dumazet
0 siblings, 0 replies; 6+ messages in thread
From: Eric Dumazet @ 2012-03-07 1:19 UTC (permalink / raw)
To: tingwei liu; +Cc: netfilter-devel
On Wed, 2012-03-07 at 08:46 +0800, tingwei liu wrote:
> On Tue, Mar 6, 2012 at 11:19 PM, Eric Dumazet <eric.dumazet@gmail.com> wrote:
> > Le mardi 06 mars 2012 à 10:05 +0800, tingwei liu a écrit :
> >> In the past few days, I have puzzled by NFQUEUE in bridge mode.
> >> I have take some test with five kernels.
> >> 2.6.24.4
> >> 2.6.36.4
> >> 2.6.38
> >> 3.0.8
> >> 3.1.10
> >>
> >> The result is : 2.6.24.4,2.6.26.4,2.6.38 have a goog performance;
> >> 3.0.8 and 3.1.10 have a poor performance.
> >> Next is copy from suricata maillist( eric@regit.org )
> >>
> >> I'm having a look at it. There has been some changes between the two
> >> kernel versions (bringing more performances) but it seems there is some
> >> side effects with bridge.
> >> Thanks!
> >
> > More data is welcomed. What is "good", what is "bad" ?
> >
> Sorry for that.
> My network is 100Mbps.
> "good" means the forward bandwidth almost 100Mbps.
> "bad" means the forward bandwidth with nfqueue only 30Mbps.
> Thanks for your reply!
more info needed... like what kind of network adapter do you use , and
ethtool -k settings, like
ethtool -k eth0
--
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: NFQUEUE in bridge mode performance poor in the last kernels
2012-03-06 21:23 ` Florian Westphal
@ 2012-03-07 1:26 ` tingwei liu
0 siblings, 0 replies; 6+ messages in thread
From: tingwei liu @ 2012-03-07 1:26 UTC (permalink / raw)
To: Florian Westphal; +Cc: netfilter-devel
On Wed, Mar 7, 2012 at 5:23 AM, Florian Westphal <fw@strlen.de> wrote:
> tingwei liu <tingw.liu@gmail.com> wrote:
>> In the past few days, I have puzzled by NFQUEUE in bridge mode.
>> I have take some test with five kernels.
>> 2.6.24.4
>> 2.6.36.4
>> 2.6.38
>> 3.0.8
>> 3.1.10
>>
>> The result is : 2.6.24.4,2.6.26.4,2.6.38 have a goog performance;
>> 3.0.8 and 3.1.10 have a poor performance.
>> Next is copy from suricata maillist( eric@regit.org )
>>
>> I'm having a look at it. There has been some changes between the two
>> kernel versions (bringing more performances) but it seems there is some
>> side effects with bridge.
>
> Might be the 'gro+nfqueue eats MAC header' problem, you could try
> commit a8db7b2d197a0d624baab83f0c810b0edbc4ffd0 (netfilter: nf_queue: fix
> queueing of bridged gro skbs). Or, disable gro on all bridge ports via
> ethtool -K $device gro off
>
I have test it.You are right.Now it is time to learn why the gro effects it.
Thanks very much.
> If its not gro related, please provide more information about your
> machine, setup, ... etc.
--
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2012-03-07 1:26 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2012-03-06 2:05 NFQUEUE in bridge mode performance poor in the last kernels tingwei liu
2012-03-06 15:19 ` Eric Dumazet
2012-03-07 0:46 ` tingwei liu
2012-03-07 1:19 ` Eric Dumazet
2012-03-06 21:23 ` Florian Westphal
2012-03-07 1:26 ` tingwei liu
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).