network drops during large file transfer

Phathead · 05-29-2006, 09:15 PM

Sorry if this has been asked before...I just don't know what I'd search on to find this.

I have a problem with eth0. It's a venerable 3com 3c905b "cyclone." I bought a bunch off eBay because they (usually) just work, and they were about $1 each in bulk.

My problem is that during a large file transfer over the wired LAN in my house, eth0 (my only network card) will quit operating. No errors in /var/log anywhere, no indication of errored packets using ifconfig. The only way I can restart it is to issue "ifconfig eth0 down" and "/etc/rc.d/rc.inet1 eth0_start" If I'm transfering at full speed via FTP, that'll work for about 10-20 seconds. I've resorted to an infinite while loop running those two commands every 20 seconds. It works--sort of, but it's a band-aid fix, and I want a real fix.

I'm running kernel 2.6.10 on Slackware 10.2. Here's the output of lspci and lsmod:

Code:

root@Nimitz:/boot# lspci
00:00.0 Host bridge: VIA Technologies, Inc. VT8366/A/7 [Apollo KT266/A/333]
00:01.0 PCI bridge: VIA Technologies, Inc. VT8366/A/7 [Apollo KT266/A/333 AGP]
00:09.0 Multimedia audio controller: Creative Labs SB Live! EMU10k1 (rev 04)
00:09.1 Input device controller: Creative Labs SB Live! MIDI/Game Port (rev 01)
00:0f.0 Ethernet controller: 3Com Corporation 3c905B 100BaseTX [Cyclone] (rev 30)
00:11.0 ISA bridge: VIA Technologies, Inc. VT8233A ISA Bridge
00:11.1 IDE interface: VIA Technologies, Inc. VT82C586A/B/VT82C686/A/B/VT823x/A/C PIPC Bus Master IDE (rev 06)
00:11.2 USB Controller: VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller (rev 23)
00:11.3 USB Controller: VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1 Controller (rev 23)
01:00.0 VGA compatible controller: ATI Technologies Inc RV280 [Radeon 9200] (rev 01)
01:00.1 Display controller: ATI Technologies Inc RV280 [Radeon 9200] (Secondary) (rev 01)

Code:

root@Nimitz:/boot# lsmod
Module                  Size  Used by
iptable_nat            23496  1
ip_conntrack           40564  1 iptable_nat
ip_tables              16768  1 iptable_nat
deflate                 2944  0
zlib_deflate           21272  1 deflate
zlib_inflate           17280  1 deflate
twofish                37376  0
serpent                13440  0
aes_i586               38644  0
blowfish                7936  0
des                    11520  0
sha256                  9088  0
w83627hf               27680  0
i2c_sensor              3072  1 w83627hf
sha1                    8448  0
crypto_null             2176  0
i2c_isa                 1920  0
xfrm4_tunnel            3204  0
ipcomp                  6792  0
esp4                    6912  0
ah4                     5632  0
af_key                 28560  0
nfsd                  202272  8
exportfs                5120  1 nfsd
lockd                  61608  2 nfsd
sunrpc                134244  2 nfsd,lockd
ipv6                  241280  30
ohci_hcd               19464  0
ehci_hcd               29828  0
pcspkr                  3432  0
joydev                  8128  0
evdev                   7552  0
via_agp                 7680  1
agpgart                28456  2 via_agp
dsbr100                 5376  0
videodev                7552  1 dsbr100
emu10k1_gp              2944  0
gameport                3712  1 emu10k1_gp
3c59x                  36648  0
via_ircc               23444  0
irda                  120764  1 via_ircc
crc_ccitt               1920  1 irda
i2c_viapro              6412  0
i2c_core               18448  4 w83627hf,i2c_sensor,i2c_isa,i2c_viapro
uhci_hcd               30352  0
fglrx                 445856  8
ide_scsi               14084  0

Like I said, no errors in /var/log anywhere, and when I run "ifconfig eth0 down" and "/etc/rc.d/rc.inet1 eth0_start" I get no error messages.

I have swapped networks cards, switches and PCI slots, and now I'm at a loss for what to do next.

This problem has been going on for months, but it started spontaneously.

Any ideas?

cwwilson721 · 05-29-2006, 11:16 PM

First: A tip to cut down on your typing:

Code:

/etc/rc.d/rc.inet1 eth0_restart

Just a bit quicker...

Second: Two pieces of info looked important:

Quote:

Originally Posted by Phathead

I'm running kernel 2.6.10 on Slackware 10.2.

And

Quote:

Originally Posted by Phathead

I have swapped networks cards, switches and PCI slots, and now I'm at a loss for what to do next.

That seems to suggest an older (2.6.10) kernel driver may be acting up, since you have eliminated hardware.

You may need to recompile the kernel with a newer version (A bit drastic, I know, but any other ideas?)

Let us know

Phathead · 06-04-2006, 12:50 AM

That is a good idea, but no joy. I compiled 2.6.16.19 today and all that did was break other things without fixing the network problem. I have so many weird modules installed, upgrading the kernel is a real ordeal, which is why I haven't done it in a year and a half. So, I'm back to 2.6.10, which mostly works.

I suspect it's some sort of motherboard problem since I've ruled out everything else.

cwwilson721 · 06-04-2006, 12:56 AM

Quote:

Originally Posted by Phathead

That is a good idea, but no joy. I compiled 2.6.16.19 today and all that did was break other things without fixing the network problem.

In other words, you didn't compile correctly. I've had to recompile 7-8 times to get everything right.

Quote:

Originally Posted by Phathead

I have so many weird modules installed, upgrading the kernel is a real ordeal, which is why I haven't done it in a year and a half. So, I'm back to 2.6.10, which mostly works.

All I have to say is try, try again. And again. And again.
Until you get all you need compiled in and modules made for your hardware.
It takes time.

Quote:

Originally Posted by Phathead

I suspect it's some sort of motherboard problem since I've ruled out everything else.

That could very well be the case.

Phathead · 06-04-2006, 01:06 AM

Quote:

Originally Posted by cwwilson721

In other words, you didn't compile correctly. I've had to recompile 7-8 times to get everything right.All I have to say is try, try again. And again. And again.

Well, yes and no. I expected that not everything would work. I need modules to make my USB FM tuner work, and I've completely lost the instructions for how to install that--assuming I could find the source. Plus, the ATI Radeon drivers would need to be uninstalled and rebuilt for the new kernel. I'm not doing that until I know the network problem is fixed by the new kernel. Also, ALSA is broken--probably because my sound card is 8 years old.

Point is, I know I could make all that work again--eventually. However, if the network problem isn't fixed with the new kernel, I'm not going to invest the time or headache to make it all work with the new kernel.