Antwort: Re: [vortex] 3c905 hanging when network load is high
torsten@thesse.de
torsten@thesse.de
Wed Jan 30 05:24:00 2002
This is a multipart message in MIME format.
--=_alternative 0038EC28C1256B51_=
Content-Type: text/plain; charset="us-ascii"
Hello,
I have read all mails to these subject because I have the same problem. My
system is a double processor machine with 2 Pentium 100 and a 3Com 5900
Combo PCI NIC. The system is SuSE Linux 7.1 with kernel 2.2.18 SMP. The
network runs with 10 MB with Half duplex. I have made the same tests as
Markus Schaub with the same results. The "fast" workstation is a SUN Sparc
10 !!!!
Additional I have tested a D-Link NIC ... and there is no change. I think,
it can be a general problem in the kernel but I don't have the know how to
search and fix these problems.
I have searched for messages like this subject in the mail database of
vortex and I have found some things about the time staying in interrupt
routine when the NIC receives a frame. Are there any parameters that I can
change and where can I find this parameters?
In addition I have the same problem with a 486 DXII/66 and ISA NIC 3Com
509 an SuSE Linux 7.1 Kernel 2.2.18, but in this machine the failure comes
not often as on the pentium machine.
Thanks for help
Torsten
On Wed, 23 Jan 2002, Markus Schaub wrote:
> maggie:# cat /dev/zero > /mnt/file
> -> timeout
> lisa:# cat /mnt/file > /dev/null
> ->timeout
x
In both these cases, lisa receives data.
> So it couldn`t be the hard disk which is `stealing' interrupts. So is
> the NIC too fast for the whole system (I made a mistake in my first
> mail: it's only a P60)?
Looks like. I don't have much experience with this, the slowest machine
that I had with PCI network cards was running at 200 MHz...
--
_______________________________________________
vortex mailing list
vortex@scyld.com
http://www.scyld.com/mailman/listinfo/vortex
--=_alternative 0038EC28C1256B51_=
Content-Type: text/html; charset="us-ascii"
<br><font size=2 face="sans-serif">Hello,</font>
<br>
<br><font size=2 face="sans-serif">I have read all mails to these subject because I have the same problem. My system is a double processor machine with 2 Pentium 100 and a 3Com 5900 Combo PCI NIC. The system is SuSE Linux 7.1 with kernel 2.2.18 SMP. The network runs with 10 MB with Half duplex. I have made the same tests as Markus Schaub with the same results. The "fast" workstation is a SUN Sparc 10 !!!!</font>
<br><font size=2 face="sans-serif">Additional I have tested a D-Link NIC ... and there is no change. I think, it can be a general problem in the kernel but I don't have the know how to search and fix these problems. </font>
<br><font size=2 face="sans-serif">I have searched for messages like this subject in the mail database of vortex and I have found some things about the time staying in interrupt routine when the NIC receives a frame. Are there any parameters that I can change and where can I find this parameters?</font>
<br><font size=2 face="sans-serif">In addition I have the same problem with a 486 DXII/66 and ISA NIC 3Com 509 an SuSE Linux 7.1 Kernel 2.2.18, but in this machine the failure comes not often as on the pentium machine.</font>
<br>
<br><font size=2 face="sans-serif">Thanks for help</font>
<br>
<br><font size=2 face="sans-serif">Torsten</font>
<br>
<br>
<br>
<br><font size=2 face="Courier New"> <br>
On Wed, 23 Jan 2002, Markus Schaub wrote:<br>
<br>
> maggie:# cat /dev/zero > /mnt/file<br>
> -> timeout<br>
<br>
> lisa:# cat /mnt/file > /dev/null<br>
> ->timeout<br>
x<br>
In both these cases, lisa receives data.<br>
<br>
> So it couldn`t be the hard disk which is `stealing' interrupts. So is<br>
> the NIC too fast for the whole system (I made a mistake in my first<br>
> mail: it's only a P60)?<br>
<br>
Looks like. I don't have much experience with this, the slowest machine <br>
that I had with PCI network cards was running at 200 MHz...<br>
<br>
-- <br>
_______________________________________________<br>
vortex mailing list<br>
vortex@scyld.com<br>
http://www.scyld.com/mailman/listinfo/vortex<br>
</font>
<br>
<br>
--=_alternative 0038EC28C1256B51_=--