lkml.org 
[lkml]   [1999]   [May]   [24]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [2.2.9] performance problem on 100MBit network
On Fri, May 21, 1999 at 07:23:12PM -0700, Simon Kirby wrote:
> On Thu, 20 May 1999, Jochen Heuer wrote:
>
> > using 2.2.9 on my server or my client brings down the performance
> > on my local network. I'm using SMC Etherpower II on both machines
> > which are connected via a 100MBit hub.
> >
> > Using ftp the transferrates drop to ~200 k/s (>6 M/s with 2.2.8).
> > If I revert the following part of patch-2.2.9 the network runs at
> > full speed again:
> >
> > diff -u --recursive --new-file v2.2.8/linux/include/net/tcp.h linux/include/net/tcp.h
> > --- v2.2.8/linux/include/net/tcp.h Tue May 11 13:10:32 1999
> > +++ linux/include/net/tcp.h Thu May 13 10:33:17 1999
> > @@ -728,7 +728,7 @@
> > */
> > extern __inline__ __u32 tcp_recalc_ssthresh(struct tcp_opt *tp)
> > {
> > - __u32 snd_wnd_packets = tp->snd_wnd / tp->mss_cache;
> > + __u32 snd_wnd_packets = tp->snd_wnd / max(tp->mss_cache, 1);
> >
> > return max(min(snd_wnd_packets, tp->snd_cwnd) >> 1, 2);
> > }
> >
> >
> > What is the reason behind this patch or can I tune my system
> > using /proc/sys/net ?
>
> This cannot be possible unless a compiler is breaking something here.
> What you are reverting avoids a possible division by zero -- the only way
> it could have been faster before was if it was spitting out divison by
> zero errors and still going. Are you changing anything else between
> kernels?

Well, I don't know what happened but at least I can not reproduce
it any more. Very strange. Plain 2.2.9 is back to full speed again :)
Maybe something else caused this slowdown. I don't know. Maybe it
has something to do with my SMC Etherpower II cards. Sometimes I get
ping times of up to 20 seconds(!) while normaly ping times are ~0.3ms.
If this happens I have to reboot either my server or my client.

What can I do to debug this if it happens again?

Regards,

Jogi

--

Well, yeah ... I suppose there's no point in getting greedy, is there?

<< Calvin & Hobbes >>

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:51    [W:0.066 / U:13.992 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site