Re: 1000ms delay in networking stack or driver, new bug?

Fumitoshi UKAI (ukai@debian.or.jp)
Tue, 21 Sep 1999 07:50:59 +0900

At Mon, 20 Sep 1999 15:48:15 -0400 (EDT),
Derek Wildstar <dwild@linux.com> wrote:
>
> On Mon, 20 Sep 1999, Arjan van de Ven wrote:
>
> > In article <199909201653.JAA00874@pizda.ninka.net> you wrote:
> >
> > > Do I sense that the common denominator in this problem is that some
> > > form of DHCP/pump/whatever is being used? If that is common amongst
> > > everyone seeing this problem, it would be nice to know.
> >
> > I only exerienced this once, so maybe it doesn't count, but:
> > I don't use DHCP, I have two NIC's though (both RTL8139) and a i386/SMP
> > machine. No special network-configuration configured. (So no pump, NAT or
> > whatever). Networkload was significant though (bursts of > 60 Mbit/s).
>
> I've experienced the same problem on an i386/single (pII) with a 3c575b
> (PCMCIA) net card, no DHCP, a several kernels and version version of
> pcmcia utils (either compiled w/gcc2.7.2.3 or gcc2.95.1/egcs 1.1.2, same
> problem either way) The NIC/PCMCIA are compiled as modules -- though I
> have yet to experience the same problem with 2.3.15-18 with cardbus
> drivers linked into the kernel.

I've experienced the same problem too, on Dual PentiumIII/500MHz
with Intel EtherExpressPro 10/100 Ethernet (10Mbps) and Digital
DS21140 (Tulip, 10Mbps) in 2.2.10 kernel, while running RealServer.
At that time, we got around 3000 ms delay by ping on DS21140 NIC.

After upgraded 2.2.12, it works fine now.

Regards,
Fumitoshi UKAI

-
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/