Ubuntu

/dev/eth0 shuts down when large file copy times out via Samba

Asked by Nateloaf

I am running Feisty amd64. I frequently copy large files (DVD Images, etc.) from a Windows XP (32 bit) box on my local network. I have no problem accessing this PC most of the time, but often when copying large files I get a timeout error and the copy fails. The provided Retry option does not work. The strange thing is that my entire network connection seems to shut off when this error occurs. I loose access to the internet, and my local network. "dmesg" shows me something about a timeout being detected and the etc1 interface being restarted (I have two NICs. eth0 is used by a VM so eth1 is my primary connection - mostly due to the order I happened to plug the cables in). Sorry I can not post the exact error as I discovered this launchpad thingie a while after it happened last.

Asus A8N32-SLI Deluxe
AMD Athlon 4400+
2GB
IDE drives

Questions: 1) Why is this happening? 2) Is there some way to force /dev/etc? to reset without rebooting?

Question information

Language:
English Edit question
Status:
Solved
For:
Ubuntu Edit question
Assignee:
No assignee Edit question
Solved by:
Andres Mujica
Solved:
Last query:
Last reply:
Revision history for this message
Nateloaf (nwild-bulk) said : 		#1

One more thing. This happens with any large transfer of data not just copying large contiguous files. If I stream music from this PC (which happens to be where my tunes live) it does the same thing regularly enough that it is unusable...

Revision history for this message
Andres Mujica (andres.mujica) said : 		#2

do you continue using your computer while the files is being transferred??

could be a power management related issue, so we need to check this part.

Thanks a lot

Revision history for this message
Nateloaf (nwild-bulk) said : 		#3

I tried copying three 4GB files over from the XP box today. I did them each one at a time and kept working on the PC while they copied. All three copied without errors.

You might be on to something. Are there power management settings somewhere that need to be fixed?

Revision history for this message
Andres Mujica (andres.mujica) said : 		#4

> You might be on to something. Are there power management settings
> somewhere that need to be fixed?

Good, please check in your BIOS settings any configuration related with Power
Management and your LAN card, could be a not desirable BIOS setting.

Revision history for this message
Alan Pope 🍺🐧🐱 🦄 (popey) said : 		#5

Nateloaf, my ASUS based desktop PC also exhibits this behavior in 32-Bit Feisty. I never quite nailed down what it might be, but I suggest you convert this support ticket to a bug as it sounds like a driver issue to me.

Revision history for this message
Nateloaf (nwild-bulk) said : 		#6

I don't think this is power related, or at least not soleley. I copied a few large files from the XP box to my Feisty box this morning, and it crapped out in the same manner, even though I was using the PC the entire time - never so much as a screensaver.

This is what dmesg gave me, not sure if it is related, but it looked promising:
[98274.148492] gnome-vfs-daemo[32151]: segfault at 0000000000000020 rip 00002ac00c09091d rsp 0000000041042040 error 4

A few more clues:
1) I am using Samba over eth1. When it dies like this, eth0 (which I use in an XP64 VM) continues to work properly.
2) I can ping the PC I am trying to access over wither interface. It would seem almost like it's just the Samba layer that is is dying if it weren't for the fact that all outgoing internet traffic seems to croak as well...

Revision history for this message
Andres Mujica (andres.mujica) said : 		#7

well i did a litle bug hunting without any success..

could you check at /var/crash for crashed logs.

Probably we would have to put a bug report for this, but i want, with your help and patience, to make a "good case" finding all the info needed from the problem.

Revision history for this message
Best Andres Mujica (andres.mujica) said : 		#8

well i digged a bit mor and found this:

https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.20/+bug/92567

seems to be a bur report already!! please check it and if you thinks it apply to you add a me too message referencing this answer #, and adding the asked info at that bug and in tis thread.

Also i've found more on google

http://forum.club.mandriva.com/viewtopic.php?p=285188&sid=3531c8ab196d692d0cde896a3e1808c2

http://forums.suselinuxsupport.de/lofiversion/index.php/t23659.html

http://forums.nvidia.com/index.php?act=Print&client=printer&f=34&t=4978

These posts seems to reflect some similar problems, if i got it wrong your board has a marvel nic and a nvidia nic, and it seems to be some sort of firewall in the nvidia nic... if this is the case, please try to disable the nvidia nic as suggested in one of the posts.

hope this helps.

Revision history for this message
Nateloaf (nwild-bulk) said : 		#9

Thanks SEAQ - Andres Mujica, that solved my question.

Revision history for this message
Nateloaf (nwild-bulk) said : 		#10

I flipped over to using the Marvel NIC and I haven't had the problem yet... Seems silly, but I'm working!

Revision history for this message
Andres Mujica (andres.mujica) said : 		#11

you're welcome. !!!