netdev
[Top] [All Lists]

Re: V2.4 policy router operates faster/better than V2.6

To: netdev@xxxxxxxxxxx
Subject: Re: V2.4 policy router operates faster/better than V2.6
From: "Jeremy M. Guthrie" <jeremy.guthrie@xxxxxxxxxx>
Date: Thu, 6 Jan 2005 16:18:21 -0600
Cc: Robert Olsson <Robert.Olsson@xxxxxxxxxxx>, Stephen Hemminger <shemminger@xxxxxxxx>
In-reply-to: <16861.46969.536236.11355@robur.slu.se>
Organization: Berbee Information Networks
References: <200501031455.26980.jeremy.guthrie@berbee.com> <200501061546.32116.jeremy.guthrie@berbee.com> <16861.46969.536236.11355@robur.slu.se>
Reply-to: jeremy.guthrie@xxxxxxxxxx
Sender: netdev-bounce@xxxxxxxxxxx
User-agent: KMail/1.7.2
On Thursday 06 January 2005 04:11 pm, Robert Olsson wrote:
> Jeremy M. Guthrie writes:
>  > >  I don't follow...
>  >
>  > Technically by turning off NAPI, I have 'solved' my short term packet
>  > loss problem.
>
>  You need NAPI of course just wait until your packet load increases just
>  a little bit... 
I agree.  Running mpstat now shows that even w/o packet drops, I'm running 
2-5% free CPU.

>  I saw drops in your 2.4 /proc/net/softnet_stat which 
>  indicates you are close to your system performance. 
Should those drops show up in RX drops in ifconfig?

>  With NAPI you keep 
>  up your system system performance regardless of incoming load.
You mentioned before that " The stats didn't show any numbers so we don't know 
your load."  Was there a command you wanted me to re-run?

>  The e1000 driver has some bugs in "your setup" as it enables irq's when
>  there is only RX and no TX and should have irq's disabled. I sent patch
>  to intel.
>
>            --ro

-- 

--------------------------------------------------
Jeremy M. Guthrie        jeremy.guthrie@xxxxxxxxxx
Senior Network Engineer        Phone: 608-298-1061
Berbee                           Fax: 608-288-3007
5520 Research Park Drive         NOC: 608-298-1102
Madison, WI 53711

Attachment: pgpZgh74fRFP6.pgp
Description: PGP signature

<Prev in Thread] Current Thread [Next in Thread>