Heavy network activity (eg: torrent/nfs file transfers) causes Hard System Locks and/or Network Freezes.

[Jonathan Strander]

Simply downloading a torrent at high speed (speed caps mitigate this somewhat) or transferring large files at high data rates over a network (for instance transfering files to NFS mounts as some users have reported) for any period of time longer than a few minutes may cause one of two conditions to occur:

1) The system Hard Locks (no i/o response from keyboard or mouse and the display manager freezes) and needs a hard reboot.
2) The network connection freezes, but still identifies itself as having a signal and being connected. The only way I could determine to fix this was to soft-reboot the system.

This is basically a complete show-stopper bug as these are common activities.

More information:
Processor: Athlon XP 2000+
#RAM: 1GB
Video: Nvidia Geforce 6600GT using driver 100.14.19
Network: DWL-520+ run using ndiswrapper
Kernel: linux-image-rt (linux-image-2.6.22-12-rt)

This occurs even when downloading to an EXT3 partition (and thus is related to networking in general and not specifically NFS). Capping torrent bandwidth will sometimes allow hours of activity of this kind, but normally it's 15 minutes or less. The system can be frozen quickly as a test by simply opening up all available bandwidth and filling it with file transfers/downloads (hence torrents causing the crash). There are three threads (possibly more) on the Gutsy Development forums.

http://ubuntuforums.org/showthread.php?t=530107
http://ubuntuforums.org/showthread.php?t=563251
http://ubuntuforums.org/showthread.php?t=563049

It would appear the common factor is wireless networking.

[Jonathan Strander (mblackwell1024)]

I have tested this with both the -rt and -generic kernels and with and without Nvidia as the display driver (in case it was some kernel module craziness). The result is the same in both, although turning off Nvidia causes the length of timeof contiguous high network activity needed before a hang to increase.

Tomorrow I will attempt to boot with an older kernel and see if I have the same result.

Beyond that I have no idea how to trace this further.

[Jonathan Strander (mblackwell1024)]

This was also tested with linux-image-lowlatency, kernel version 2.6.20-16, and this crash did not occur.

[Deepak Manoharan (deepakm+ubuntu)]

I see the same issue, Its not related to wireless though in my case. A heavy udp burst using iperf on a wired network causes gutsy to become unresponsive. To get it back to the old stage, I had to stop the traffic coming through and takes a couple of minutes to become alright.

I dont think I have had this issue with 7.04 and dont think I can try with it either.

-Deepak

[Stavros Korokithakis (stavrosk)]

I can confirm this. I have a server that backs up files from other computers to a NAS device (it retrieves the files through CIFS and stores on the NAS through NFS). After a bit of heavy copying, the ethernet adapter locks up (the system is responsive, though I'm not sure because it's headless, I just see log entries), but the ethernet device will not accept or emit packets. It's a pretty serious bug. I worked around it by setting large packet sizes and it worked for a bit, but now it's back to broken.

[Chow Loong Jin (hyperair)]

Okay, so thank goodness I'm not the only one. I was already considering a reinstall, because I couldn't find anything related to my bug for days. If it helps, the syslog doesn't show anything related to the crash, though at one point of time: I had this assertion failure.

Oct 12 16:29:03 Hyperair-PC kernel: [ 1185.457745] KERNEL: assertion (!atomic_read(&sk->sk_wmem_alloc)) failed at /build/buildd/linux-source-2.6.22-2.6.22/net/ipv4/af_inet.c (149)

[Chow Loong Jin (hyperair)]

Something that might be worth highlighting is that ndiswrapper (version 1.45, which is included in the current Ubuntu kernel 2.6.22-14-generic) has a bug whereby it crashes with large transfers. This has been fixed in version 1.46, and the current version of ndiswrapper is 1.48. Perhaps this version of ndiswrapper may be included into the next version of the kernel?

More info: http://ndiswrapper.sourceforge.net/joomla/index.php?/content/view/16/2/

[Stavros Korokithakis (stavrosk)]

I have installed gutsy server on a new pc and am not getting this problem any more. I upgraded the old pc from edgy to feisty and there was an update for nfs-utils (I think that's the name), but I haven't tested to see whether it is fixed there.

[undine (niirti)]

I can confirm that the same bug on my Travelmate 4402WLMI. I am also using Ndiswrapper for my BCM4318.

[Chow Loong Jin (hyperair)]

Okay, so I tried using the 4.18 ndiswrapper driver, but the same error still happened. So I'm using the ACX driver for my DWL-520+ I'll confirm tomorrow if the lockup still occurs.

[Chow Loong Jin (hyperair)]

Make that 1.48, I mistyped it ><

[sanjuan (eric-sanjuan)]

Hi,
I'm not sure it comes only from ndiswrapper,
This module is not installed on a new dell dual core 2 server (no graphic interface)
with gutsy Linux 2.6.22-9-generic #1 x86_64 GNU/Linux installed
and I'm facing the same troubles at least once a week.

The netwokf interface is freezing, the server only responds to ping.
What it is amazing is that when I wait some hours (on or two)
I can connect again normally to the computer and there is no trace
of the freezing in the log files.

At the beegenning the problem happenned daily each 24 hours,
I stopped the cron tables to see if that was the problem,
the netwok still freezing but less often.

Any idea if my problem is related to yours?

[Chow Loong Jin (hyperair)]

I have no idea, really. It only happens with torrents on my computer, and never otherwise. But then again, my maximum throughput is 70kB/s. I hate my ISP. You say the problem happens every 24 hours. I noticed that every 24 hours, apt-check is run, and that does generate quite a bit of network traffic. Could that be it?

My log files also show no trace of the error. It just shows the time i restart. Which means that the kernel is not completely frozen. syslogd still runs. The magic keys refuse to work though.

[Chow Loong Jin (hyperair)]

On a side note, I noticed that users of other distros have similar problems, and the common factor seems to be ndiswrapper.

[Stavros Korokithakis (stavrosk)]

Not so with me, I was using Ubuntu's native drivers.

[sanjuan (eric-sanjuan)]

neither with me, I'm not using ndiswrapper but it could be a similar problem with Ubuntu's native drivers and some recent network cards.

I agree with the idea that apt-get could cause such freezing because of the network trafic it generates.

[Chow Loong Jin (hyperair)]

Strange eh, how about everyone posting their network card models and the drivers used here so the developers can work something out?

[Chow Loong Jin (hyperair)]

Okayy, so I tried downloading a torrent last night using the acx module instead of the ndiswrapper module. No crash. I guess I'll be using that for heavy traffic =\

[jasongegere (jason-hgmail)]

I have been doing some research trying to figure out why my installation of Feisty 7.0.4 locks up and I ultimately found this bug report. I can confirm many of the items that are talked about in this report. About or exactly at 2gb of network traffic using SMB (samba) the system locks up, kernel panic with a beep and the numlock and scolllock flashing a few times. I also am able to make the system lock up while trying to create a new disk with VMware Server. At around 2gb the system also locks up. I am running a Intel Xeon 2.6 64bit on a Asus motherboard.

The weird thing is I was able to download a 3.5gb Fedora Core 5 DVD image with Firefox.

Hope this helps. I would love to find a solution.

[Chow Loong Jin (hyperair)]

I never had a problem on Ubuntu Feisty, and certainly this problem only happens for me on Gutsy with ndiswrapper. I can't figure out why though. It seems that it was a bug to do with version 1.45, but even after installing 1.48 I still had the problem.

[Adolfo González Blázquez (infinito)]

Could this be the same bug as https://bugs.launchpad.net/ubuntu/+source/ndiswrapper/+bug/136814 ??

[Chow Loong Jin (hyperair)]

Ah thanks for referring me there. That's definitely my bug.

[Matthew Specker (matthew-specker)]

I've been experiencing the same problem running Gutsy for the last month or so. Only occurs during large transfers over NFS. No ndiswrapper here. I'm running an Intel wireless card.

[effell (effell)]

I'm experiencing the same problem on a PCMCIA 3G/HSDPA wireless card (a Novatel Merlin) on a Thinkpad X31.
High bandwidth (eg bittorrent) quickly causes a system freeze.
I'm available for performing further testing, in case anyone needs.

[Chow Loong Jin (hyperair)]

Are you running via ndiswrapper? If so go to http://ndiswrapper.sourceforge.net and go download the 1.49rc4 version. That will fix the bug. I've been running for days now on that.

[Chow Loong Jin (hyperair)]

Sorry I meant 1.49. I didn't realize it had already been released. And launchpad.net should allow editing of posts!

[effell (effell)]

Not running NDISwrapper, not even a wifi card. This is a 3G cellular data card, showing up on a couple of ttyUSB ports courtesy of the kernel Option module, it seems. From then on it accepts Hayes AT commands like a regular modem.

[Chow Loong Jin (hyperair)]

Well it might be then that I don't get this bug because my maximum download/upload speeds for my internet connection respectively are: 512kbps/256kbps or 64kBps/32kBps. Miserable huh.

[effell (effell)]

For the record, in my case the system locks up at or below 80kBps.
Running FC7 on the same card/network (but different Thinkpad) has been fine.

[Chow Loong Jin (hyperair)]

Well why don't you get an Ubuntu Gutsy LiveCD and try doing some heavy transfers on that other Thinkpad to see if the problem persists?

You could also try using an FC7 LiveCD on the Thinkpad that locks up and see if the problem persists to be sure if it's really Ubuntu's fault.

[effell (effell)]

Just to clarify the above, I didn't mean that no lockup occurs above 80kBps, only that it occurs even at speeds (as low?) as below 80.

The hardware seems fine: ran self-test on CPU, ran memtest, network card runs fine on another laptop. The lock-up occurs only under heavy network traffic, so I'm leaning towards a software cause and I think some sort of postmortem info would be the most useful. I'll go and find out how to obtain that, but would appreciate pointers on how to do it in this specific circumstance, ie, full lock-up.

Still, for kicks I'll also add further data points by running live CDs later on.

[JackC (jack-jncsoftware)]

I'm experiencing the same issue. Large file copies will hard crash the machine in a matter of minutes. This is over a wired connection. No wireless card is even installed.

[-emory- (emory-roane)]

AHA! I think I solved the problem. What I think was happening was that gutsy wasn't using swap correctly. Change swappiness to 10 and it works beautifully so far!
btw to do so just do
sudo gedit /etc/sysctl.conf
then add the line swappiness=10 to the bottom, if it isn't already there. Can anyone confirm this for me?

[-emory- (emory-roane)]

ouch...looks like i spoke a bit too soon. All worked well until I loaded amarok. It could have been a coincidence of course that it started to crash as soon as that loaded up, I'm not sure. Any luck on anyone else? I wonder why everyone isnt experiencing this though...

[effell (effell)]

A long shot: here's a possible extra clue for solving this problem. Twice after freezing this way and being left alone for a while, my Thinkpad X31 would not even turn on with the battery plugged in -- it was necessary to remove it. Some sort CMOS corruption connection?

Anyway, again if anyone working on this wants specific trace info or some test performed, please do ask.

[jasongegere (jason-hgmail)]

I had posted earlier in this forum. My system locked ups were also on
a wired connection.

--
HTMLgraphic Designs, LLC
[phone] x101
www.htmlgraphic.com

On Nov 3, 2007, at 10:45 PM, JackC <email address hidden> wrote:

> I'm experiencing the same issue. Large file copies will hard crash the
> machine in a matter of minutes. This is over a wired connection. No
> wireless card is even installed.
>
> --
> Heavy network activity (eg: torrent/nfs file transfers) causes Hard
> System Locks and/or Network Freezes.
> https://bugs.launchpad.net/bugs/147464
> You received this bug notification because you are a direct subscriber
> of the bug.

[irwjager (jager49)]

Vostro 1400 and intel 3945 wireless. Not using NDIS wrapper. System consistently locks up after +/- 30 minutes of bittorrent traffic (+/- 100kb upload, +/-40 kb download, single torrent).
No traces in any system log or ring buffer of a crash or problem. System completely freezes; mouse cursor just stops moving. System monitor shows no unusual activity up until freeze.

I'm not sure what to post here to help developers...

[Chow Loong Jin (hyperair)]

It occurred to me that the problem could be the network stack hanging. I mean X does depend on the network stack doesn't it? What happens if you're logged into a terminal instead? Would it still lock you out? What about processes in the background? A way to find out might be to run a program that writes to a file the time, at every, say, 5-second interval, for example:

#!/usr/bin/perl
while (1)
{
open(FILE,"$ENV{HOME}/test");
print FILE time();
close(FILE);
sleep(5);
}

And running that in the background.

[Tareeq (spock-rock)]

I would like to point out that I have encountered this as well, it seems the error occurs when using torrents while streaming over a windows share with samba. This pretty much hardlocks the machine, and syslog has nothing in it. I will post my syslog the next time it happens.

Tareeq

[JackC (jack-jncsoftware)]

It appears the problem is either with the 64 bit version on something on the desktop install. I was running the 64-bit desktop install and getting crashes every few minutes of heavy network use. I installed the 32-bit server on the same machine and transfered well over 50GB last night without any problems.

[-emory- (emory-roane)]

are we sure it's with the 64 bit version? I'm running 32 bit desktop and it's doing this... This is the ONLY thing that I've noticed wrong with this release too... In other news, I just did a fresh install and its' still messing up. I couldn't test on teh live cd though. Has anyone tried with the alternate install disk?

[JackC (jack-jncsoftware)]

I didn't know whether it was the 64 bit or the desktop since I changed both at the same time. If it crashes for you on 32 bit desktop that seems to point to the desktop.

[Jonathan Strander (mblackwell1024)]

Added system information logs.

Also note that this freeze completely locks all IO, making the keyboard non-functional including SysRq commands.

[nox.freak (nox-freak)]

Seems to be a memory leak, I can watch my system's memory slowly get eaten away tell it reaches 100% and I get a kernel panic. If I kill any applications that are transmitting or receiving over the network then memory usage levels out but dose not go down.

[nox.freak (nox-freak)]

last lock up memory was not at 100%, but memory usage was still slowly being eaten away.

SYSINFO:
AMD64 3200
1GB Ram
Ubuntu 7.10 i386
2.6.22-14-generic
NVidia GF6200
Wireless Network

[shearn89 (shearn89)]

I'd like to add that i have this same problem - using ndiswrapper 1.48 on kernel 2.6.22-14 (gutsy). I've booted to 2.6.22-12, and as yet no problem. Managed to download the entire Archlinux iso using rtorrent with no problem.

info:
Duron 640mhz proc
512mb ram
ubuntu 7.10 i386
2.6.22-14 (trying -12 to see if that changes anything)
ndiswrapper 1.48 with linksys drivers.
onboard graphics.

[Chow Loong Jin (hyperair)]

shearn89, I think your bug does not belong here, but is the ndiswrapper bug. Why don't you try installing ndiswrapper 1.49, now that it's out, on your 2.6.22-14 kernel and trying it out? I've never had problems since using 1.49.

[-emory- (emory-roane)]

Just curious, is everyone on a fresh install of gutsy, or are these problems also happening upgrading from feisty as well? Also, shearn89 and hyperair, to me that sounds like the same issue to me. Did you downgrade your kernel to get this to work? I'm about five minutes from installing Feisty back on this box.

[Andre Ruiz (andre-ruiz)]

Hi all. I can confirm this bug.

Fresh install of Gutsy 64 in a AMD X2 4400+ / Asus motherboard with builtin Gigabit ethernet (nvidia chipset). No wireless involved, just gigabit ethernet. My home is mounted over nfs, so is many directories on the machine.

I dist-upgraded just after installing, so I don't know if it would help to use old packages (like kernel, etc.).

Every time I copy a big amount of data between NFS mounts, the machine locks up. I can move the mouse and change consoles using ALT+CTRL+Fn, but X becomes unresponsive (I think this may be because my home is NFS mounted). In text consoles, I can do pretty much except try to access the network, which will hang also.

This is very reproduceable, just move a ISO around (the network) and everything freezes. The NFS server is the same as before, I'm pretty sure it's not the problem.

[effell (effell)]

Since emory asked, I'm experiencing this bug on a system upgraded from Feisty.

[-emory- (emory-roane)]

Well I just downgraded to Feisty and I don't have the problem. One thing I
just realized though, the rtl-wifi drivers that are now included in gutsy
were not included in feisty, and have to be manually installed. Perhaps this
is the root of the problem, a new version of the rtl-wifi drivers that is.

On Nov 12, 2007 6:06 PM, effell <email address hidden> wrote:

> Since emory asked, I'm experiencing this bug on a system upgraded from
> Feisty.
>
> --
> Heavy network activity (eg: torrent/nfs file transfers) causes Hard System
> Locks and/or Network Freezes.
> https://bugs.launchpad.net/bugs/147464
> You received this bug notification because you are a direct subscriber
> of the bug.
>

--
emory

[Jonathan Strander (mblackwell1024)]

That would only make sense if everyone here was using Realtek wireless cards. As you can see though, I (for instance) am using a D-Link (ACX module) card, and many of us (although not all) are using Ndiswrapper as well.

The fact that it can bring down the entire machine, forcing a hard reset, and that it doesn't happen with older kernel versions would seem to point to a kernel bug.

[nox.freak (nox-freak)]

After upgrading my ndiswrapper to v1.49 the kernel panics were
replaced with my wireless network connectivity locking up and
requiring a reboot to connect agen. I am still seeing my memory being
eaten up as well.

On Nov 12, 2007 3:39 PM, Jonathan Smith <email address hidden> wrote:
> That would only make sense if everyone here was using Realtek wireless
> cards. As you can see though, I (for instance) am using a D-Link (ACX
> module) card, and many of us (although not all) are using Ndiswrapper as
> well.
>
> The fact that it can bring down the entire machine, forcing a hard
> reset, and that it doesn't happen with older kernel versions would seem
> to point to a kernel bug.
>
> --
>
> Heavy network activity (eg: torrent/nfs file transfers) causes Hard System Locks and/or Network Freezes.
> https://bugs.launchpad.net/bugs/147464
> You received this bug notification because you are a direct subscriber
> of the bug.
>

--
Thank you,

+================+
 Winter Faulk
 <email address hidden>
 http://down-time.com
+================+

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7

iQEVAwUBRxViGvtG09nU1dFHAQI9Uwf8Do85dCiVnd2DZZJ62wFOAVd5SNutZYfh
3Bv1UyufMP+b8VmTNvQkjmefdnQNAPa+EG1jAZ4Crd1GWSu/2cVriuL+MfV29sGq
OwmwHu9ULfxU+vF8+GH20CvFVR/TbUhzPtYnMGOrL3svbvoMQ0mg9tJ/xc+MWvWd
OH9cSvDBZa29L3UNU3MM/UDx+PXY1rOYQcL1fNXLMrpgudMFBy3itbxyco10mFS+
xJLOqdzL2iEJ+Dje1tkMdVM4ilT9tO/FFI8bjhU8SlIqZFqhj3Q2+uHnUzn9ZFn0
Qtj+hr9OB1BBXK+7bTPXlMdzB7JevJ88VUtewvi3OcAX0IDhv81xSA==
=OY5j
-----END PGP SIGNATURE-----

[Chow Loong Jin (hyperair)]

Emory, this bug happened to me both on an upgraded system and a fresh install. Reverting back to Feisty's kernel solves the problem. Upgrading to ndiswrapper 1.49 also solves the problem for me.

nox.freak, have you tried unloading the ndiswrapper module, waiting for some time and reloading it? Also could you check your syslog for Ndiswrapper related entries?

To everyone using ndiswrapper, could you check if your bug is actually https://bugs.launchpad.net/ubuntu/+source/ndiswrapper/+bug/136814?

I believe that there are really two separate bugs being discussed here. One being the Ndiswrapper bug, and one being the network bug in general.

[Andre Ruiz (andre-ruiz)]

I am pretty sure they are two different bugs. At one side, people with ndiswrapper problem are having kernel panics, on the other side my machine is hanging without using wireless at all.

How do you know which comments are from ubuntu/canonical staff? I see many users discussing the bugs but it's not obvious if anyone is actually working on it.

[Chow Loong Jin (hyperair)]

Thing is, this page comes up higher on Google, compared to the ndiswrapper bug. Perhaps we should direct the people with ndiswrapper bugs to the other bug.

As for identifying the staff, I also don't know how to identify them to be honest.

[shearn89 (shearn89)]

I upgraded to v1.49 of ndiswrapper, and it seems to solve the problem - it use to crash when streaming to iTunes, and no longer does. Anyone that is using ndiswrapper and comes to this page should try upgrading first.

[Chow Loong Jin (hyperair)]

Definitely. I still don't know why Ubuntu didn't include the new version of Ndiswrapper at its release. It used 1.45, when at that time 1.48 stable was already out.

[Jonathan Strander (mblackwell1024)]

Ndiswrapper v1.49 seems to fix the torrent/multiple large downs = crash problem. (so I'm confirming that as a fix)

I'm not sure about NFS however, I haven't done any tests to see if there's still an additional underlying issue.

[Chow Loong Jin (hyperair)]

I'd like to ask those who are experiencing the problem (not with ndiswrapper) to verify if it only happens with NFS, or if it also happens with other forms of network transfers?

I've used Avahi to connect to my sister's work laptop running Windows XP, and she downloaded files from my HTTP server at speeds in the range 5-7 MB/s. I never experienced any lockups.

[Andre Ruiz (andre-ruiz)]

Now that I'm thinking of it, it may be nfs only, indeed. In the last days I have downloaded many ISO files and I do not remember the machine freezing not even once while downloading. Then, after, when copying these files to my server (nfs mounted), the locks ocurred. I will do more tests to confirm that and post back.

[-emory- (emory-roane)]

well i just downgraded back to feisty... but with me the issue was
definitely downloading torrents. And I don't use ndiswrapper...

On Nov 14, 2007 7:14 AM, Andre Ruiz <email address hidden> wrote:

> Now that I'm thinking of it, it may be nfs only, indeed. In the last
> days I have downloaded many ISO files and I do not remember the machine
> freezing not even once while downloading. Then, after, when copying
> these files to my server (nfs mounted), the locks ocurred. I will do
> more tests to confirm that and post back.
>
> --
> Heavy network activity (eg: torrent/nfs file transfers) causes Hard System
> Locks and/or Network Freezes.
> https://bugs.launchpad.net/bugs/147464
> You received this bug notification because you are a direct subscriber
> of the bug.
>

--
emory

[Chow Loong Jin (hyperair)]

So it's not just NFS transfers eh? At what speeds do your torrent
transfers usually go, and which torrent client do you use? Also, how
long before the system locks up?

-emory- wrote:
> well i just downgraded back to feisty... but with me the issue was
> definitely downloading torrents. And I don't use ndiswrapper...
>
> On Nov 14, 2007 7:14 AM, Andre Ruiz <email address hidden> wrote:
>
>
>> Now that I'm thinking of it, it may be nfs only, indeed. In the last
>> days I have downloaded many ISO files and I do not remember the machine
>> freezing not even once while downloading. Then, after, when copying
>> these files to my server (nfs mounted), the locks ocurred. I will do
>> more tests to confirm that and post back.
>>
>> --
>> Heavy network activity (eg: torrent/nfs file transfers) causes Hard System
>> Locks and/or Network Freezes.
>> https://bugs.launchpad.net/bugs/147464
>> You received this bug notification because you are a direct subscriber
>> of the bug.
>>
>>
>
>
>

[-emory- (emory-roane)]

the actual speed where it crashes veries, but usually it's about only 20kbs.
I've used all torrent clients available... literally. As soon as it starts
getting any speed at all the system hikes up to 100 percent and then stays
that way for anywhere between 5 seconds and 15 minutes, but it always
crashes...

On Nov 14, 2007 8:44 AM, hyperair <email address hidden> wrote:

> So it's not just NFS transfers eh? At what speeds do your torrent
> transfers usually go, and which torrent client do you use? Also, how
> long before the system locks up?
>
> -emory- wrote:
> > well i just downgraded back to feisty... but with me the issue was
> > definitely downloading torrents. And I don't use ndiswrapper...
> >
> > On Nov 14, 2007 7:14 AM, Andre Ruiz <email address hidden> wrote:
> >
> >
> >> Now that I'm thinking of it, it may be nfs only, indeed. In the last
> >> days I have downloaded many ISO files and I do not remember the machine
> >> freezing not even once while downloading. Then, after, when copying
> >> these files to my server (nfs mounted), the locks ocurred. I will do
> >> more tests to confirm that and post back.
> >>
> >> --
> >> Heavy network activity (eg: torrent/nfs file transfers) causes Hard
> System
> >> Locks and/or Network Freezes.
> >> https://bugs.launchpad.net/bugs/147464
> >> You received this bug notification because you are a direct subscriber
> >> of the bug.
> >>
> >>
> >
> >
> >
>
> --
> Heavy network activity (eg: torrent/nfs file transfers) causes Hard System
> Locks and/or Network Freezes.
> https://bugs.launchpad.net/bugs/147464
> You received this bug notification because you are a direct subscriber
> of the bug.
>

--
emory

[Chow Loong Jin (hyperair)]

Okay then I'm definitely not having that problem, and nor are my friends.
All of us have speeds varying 50-100 kB/s with torrents and none of us have
reported any crashes

On Nov 15, 2007 2:23 AM, -emory- <email address hidden> wrote:

> the actual speed where it crashes veries, but usually it's about only
> 20kbs.
> I've used all torrent clients available... literally. As soon as it starts
> getting any speed at all the system hikes up to 100 percent and then stays
> that way for anywhere between 5 seconds and 15 minutes, but it always
> crashes...
>
> On Nov 14, 2007 8:44 AM, hyperair <email address hidden> wrote:
>
> > So it's not just NFS transfers eh? At what speeds do your torrent
> > transfers usually go, and which torrent client do you use? Also, how
> > long before the system locks up?
> >
> > -emory- wrote:
> > > well i just downgraded back to feisty... but with me the issue was
> > > definitely downloading torrents. And I don't use ndiswrapper...
> > >
> > > On Nov 14, 2007 7:14 AM, Andre Ruiz <email address hidden> wrote:
> > >
> > >
> > >> Now that I'm thinking of it, it may be nfs only, indeed. In the last
> > >> days I have downloaded many ISO files and I do not remember the
> machine
> > >> freezing not even once while downloading. Then, after, when copying
> > >> these files to my server (nfs mounted), the locks ocurred. I will do
> > >> more tests to confirm that and post back.
> > >>
> > >> --
> > >> Heavy network activity (eg: torrent/nfs file transfers) causes Hard
> > System
> > >> Locks and/or Network Freezes.
> > >> https://bugs.launchpad.net/bugs/147464
> > >> You received this bug notification because you are a direct
> subscriber
> > >> of the bug.
> > >>
> > >>
> > >
> > >
> > >
> >
> > --
> > Heavy network activity (eg: torrent/nfs file transfers) causes Hard
> System
> > Locks and/or Network Freezes.
> > https://bugs.launchpad.net/bugs/147464
> > You received this bug notification because you are a direct subscriber
> > of the bug.
> >
>
>
> --
> emory
>
> --
> Heavy network activity (eg: torrent/nfs file transfers) causes Hard System
> Locks and/or Network Freezes.
> https://bugs.launchpad.net/bugs/147464
> You received this bug notification because you are a direct subscriber
> of the bug.
>

--
Hyperair

[Tareeq (spock-rock)]

Mine seems only to happen during torrrent transfers and streaming to my media box. I thought I would check out my memory but memtest over 9+hours showed I did not get one single error over that time period.

Nov 16 03:05:56 localhost nmbd[6142]: [2007/11/16 03:05:56, 0] nmbd/nmbd_browsesync.c:find_domain_master_name_query_fail(351)
Nov 16 03:05:56 localhost nmbd[6142]: find_domain_master_name_query_fail:
Nov 16 03:05:56 localhost nmbd[6142]: Unable to find the Domain Master Browser name CHESSCLUB<1b> for the workgroup CHESSCLUB.
Nov 16 03:05:56 localhost nmbd[6142]: Unable to sync browse lists in this workgroup.
Nov 16 03:05:56 localhost nmbd[6142]: [2007/11/16 03:05:56, 0] nmbd/nmbd_browsesync.c:find_domain_master_name_query_fail(351)
Nov 16 03:05:56 localhost nmbd[6142]: find_domain_master_name_query_fail:
Nov 16 03:05:56 localhost nmbd[6142]: Unable to find the Domain Master Browser name CHESSCLUB<1b> for the workgroup CHESSCLUB.
Nov 16 03:05:56 localhost nmbd[6142]: Unable to sync browse lists in this workgroup.
Nov 16 03:07:00 localhost ntop[6004]: **WARNING** packet truncated (16114->8232)
Nov 16 03:09:30 localhost ntop[6004]: **WARNING** packet truncated (9166->8232)
Nov 16 03:10:08 localhost ntop[6004]: **WARNING** packet truncated (14654->8232)
Nov 16 03:10:08 localhost ntop[6004]: **WARNING** packet truncated (10274->8232)
Nov 16 19:48:24 localhost syslogd 1.4.1#21ubuntu3: restart.
Nov 16 19:48:24 localhost kernel: Inspecting /boot/System.map-2.6.22-14-generic

That is the last thing I see in my syslog, as you can see 3am my system locked up became unresponsive, and I restarted the machine at 7:48pm. I am using my NF4 gigabit nic. I could try switching the nic to see if that makes a difference.

Tareeq

[thechef (d1m1)]

I had a full system freeze (can't use mouse, keys, no network response / pong) on 2.6.23 and 2.6.22.10 when I was transferring data. The freeze only occured after a few days of runtime and only when transferring data. Not sure if it was always from or to NFS drives, but there were always NFS drives mounted. The last time it crashed was when I was transferring about 70 GB over the network from an NFS drive to another NFS drive. As I really have difficulty to reproduce this problem I ask if anyone has a system on which it is easier to reproduce "this"(assuming that there is only one error) problem. Can you test the newest kernel versions?
>=2.6.23.8

Is it possible that an error such as this, can cause the problem? It's just the one that looks most likely related, but I don't know what happens to a infinite loop under the big kernel lock and how probable it is, under the known conditions, that the kernel enters this infinite loop. (I'm not a kernel programmer after all)
http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.23.y.git;a=commit;h=97855b49b6bac0bd25f16b017883634d13591d00

[Chow Loong Jin (hyperair)]

Fascinating. So these kernel freezes are actually caused by the kernel
entering an infinite loop? We should also check with other distros to
see if the problem occurs with them. I know that none of the Ubuntu
users I know personally seem to have the problem though. Including me.

thechef wrote:
> I had a full system freeze (can't use mouse, keys, no network response / pong) on 2.6.23 and 2.6.22.10 when I was transferring data. The freeze only occured after a few days of runtime and only when transferring data. Not sure if it was always from or to NFS drives, but there were always NFS drives mounted. The last time it crashed was when I was transferring about 70 GB over the network from an NFS drive to another NFS drive. As I really have difficulty to reproduce this problem I ask if anyone has a system on which it is easier to reproduce "this"(assuming that there is only one error) problem. Can you test the newest kernel versions?
>
>> =2.6.23.8
>>
>
> Is it possible that an error such as this, can cause the problem? It's just the one that looks most likely related, but I don't know what happens to a infinite loop under the big kernel lock and how probable it is, under the known conditions, that the kernel enters this infinite loop. (I'm not a kernel programmer after all)
> http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.23.y.git;a=commit;h=97855b49b6bac0bd25f16b017883634d13591d00
>
>

[thechef (d1m1)]

Oh btw. the freeze occurs here on debian etch and i compiled the kernel myself using make-kpkg based on the 2.6.18 config file (near default, but: 1000hz (if that matters on a dynticks, tickless system, i don't know), Low Latency Desktop and Preempt BKL).
I have to use the noapic boot option, because the system won't boot otherwise. (That's for all kernels the case)

[Matthew Woerly (nattgew)]

bcm43xx driver, 4306 chipset

Nothing found except the attached in syslog after a crash. I was downloading a torrent, which was doing fine, and then tried to send an iso to another computer via nfs over the network, and it froze.

Later, when just moving the iso, the transfer would go very quickly, for a second or two, stop for a few seconds, and then back to quickly and so on. The system also froze up a few times for a few seconds before continuing normally. I couldn't find anything in the logs.

[Sébastien (sesteve)]

It happens to me on a Gutsy server upgraded from feisty. It is used as a firewall and NFS file server for the lan.

Lock on happens:
* when transferring files from an NFS share to another one on the same computer from a client
* when I'm downloading torrents after a while (from 2hours to 12)

I thought it was a NFS issue because (1) happened to me first, and my torrent files (2) were located on one NFS share, but I tried moving everything on my main desktop.

Firewall locks itself on after a while, exactly same symptoms. Nothing could be spotted in syslog. So:
* lot of IO on network

Only solution I found was to hardreboot. No link with ndiswrapper as everything is using ethernet.

Nothing happens with activities like:
* NFS transfers from server to client (or from client to server)
* basic networking (web surfing, mail, ssh, apt-get....)

Basically, if I do not download torrents or copy files from a NFS share to a NFS share it works like a charm for days.

It looks my file server is like my fellow french railway civil servants.... on strike and fucking my day as I just want to work :)

Regards,
Sebastien

[thechef (d1m1)]

I'm not having any trouble since kernel 2.6.23.8, 2.6.23.9, but since the bug doesn't appear very often on my system, it's not appropriate for testing and making any more or less reliable conclusions.

Please update your kernel and test it

On debian-based systems i'd update as follows:
Download the sources from kernel.org and extract them

as root:

cd linux-2.6.23.9
cp /boot/config-2.6.22* .config #Copy the config of your current directory to the linux source directory and name it ".config"
apt-get install kernel-package #Not sure if more is required
make menuconfig #I think you can just exit after, if you don't want a configuration that is similar to your old one
make-kpkg --initrd kernel_image
make-kpkg --initrd kernel_headers #Required if you want to install proprietary graphics drivers and such later

Now there should be 2 newly created deb-packages in the parent directory of the linux source directory. Install them and reboot.

[thechef (d1m1)]

Oops, correction:
make-kpkg kernel_headers

the argument "--initrd" is not required if you create the kernel headers

[-emory- (emory-roane)]

well I finally wired hte house and now that i'm on the wired connection the
problem no longer exists. my problem is solved!

On Dec 1, 2007 5:10 PM, thechef <email address hidden> wrote:

> Oops, correction:
> make-kpkg kernel_headers
>
> the argument "--initrd" is not required if you create the kernel headers
>
> --
> Heavy network activity (eg: torrent/nfs file transfers) causes Hard System
> Locks and/or Network Freezes.
> https://bugs.launchpad.net/bugs/147464
> You received this bug notification because you are a direct subscriber
> of the bug.
>

--
emory

[Chow Loong Jin (hyperair)]

Don't forget:
make-kpkg --initrd modules_image

For users of proprietary blobs.

[Henrik Nilsen Omma (henrik)]

This bug was nominated for Gutsy but does currently not qualify for a 7.10 stable release update (SRU) and the nomination is therefore declined.
According the the SRU policy, the fix should already be deployed and tested in the current development version before an update to the stable releases will be considered. With 7.10 now released, that policy applies to this bug. See: https://wiki.ubuntu.com/StableReleaseUpdates .
The bug is not being closed as work will continue on fixing it for the next release, Hardy Heron (8.04). If the state of this bug should change such that it qualifies for the SRU process, please contact the person who originally declined it and ask them to re-evaluate it. To help improve the state of this bug see: https://wiki.ubuntu.com/Bugs/HowToTriage .

[effell (effell)]

Updating the kernel from 2.6.22-14-generic to 2.6.23.9 seems to have eliminated this bug from my system.

I followed instructions from http://ubuntuforums.org/showthread.php?t=311158

[Chow Loong Jin (hyperair)]

That's definitely good news. That means Hardy won't have a problem!

On Dec 7, 2007 11:24 PM, effell <email address hidden> wrote:

> Updating the kernel from 2.6.22-14-generic to 2.6.23.9 seems to have
> eliminated this bug from my system.
>
> I followed instructions from
> http://ubuntuforums.org/showthread.php?t=311158
>
> --
> Heavy network activity (eg: torrent/nfs file transfers) causes Hard System
> Locks and/or Network Freezes.
> https://bugs.launchpad.net/bugs/147464
> You received this bug notification because you are a direct subscriber
> of the bug.
>

--
Hyperair

[Sean Corbett (seanbutnotheard)]

Hello,

I'm running the latest Hardy beta (kernel 2.6.24-16-generic) and I'm having this issue also. For me it happens when I do a large file transfer to either an SMB share or via FTP. It's hard to say whether it's the network or disk activity that triggers it. The system locks hard... a reset is necessary. It won't respond to the keyboard or network requests.

Unfortunately the syslog doesn't give any useful information at the time of the freeze, though it does log the restart... Anyone have any tips on tracking down the cause?

[Wolfgang Glas (wglas)]

Well, this hard lock is observed by us in *every* ubuntu kernel starting with feisty's 2.6.20 flavor. We observed hard locks on webservers and on our laptops. The common factor seems to be a large number of simultaneously opened TCP/IP connections either through the number of clients in the case of the web server of through a low-bandwidth internet connection (GPRS/UMTS) on the laptop. (With slow bandwidth I tend to open multiple tabs on firefox and then have a cup of coffee...)

[Anthony Chianese (achianese)]

I'm having a problem that appears to be the same as this, also using the latest Hardy beta. I'm using network-manager, which automatically detected my Linksys WMP54G wireless PCI card as "RaLink RT2561/RT61 802.11g PCI". It works very well until the crash, which causes the system to become completely unresponsive (no keyboard or mouse response, Capslock and Scroll lock blink, nothing is added to syslog at the time of crash).

I can reproduce the crash easily by asking rsync to sync a directory across my local network.. After 2-3 minutes, it crashes. I'm willing to provide any additional information someone wants, but I'm not sure right how to get more information about the cause. The system has also crashed while I was surfing the web using firefox or downloading a large number of emails in thunderbird, always while the network is transferring data.

[Martijn Faassen (faassen)]

I just upgraded to hardy and noticed this (or a related) problem on a laptop where I didn't have this problem before on gutsy. I haven't seen a crash yet, but when I transfer a large file (using rsync or scp) to the machine, the network stops functioning (can't browse anymore, ping, etc). In case of ping I've also noticed pings with a huge latency. The large file transfer does continue but blocks all network traffic. As soon as I stop the large file transfer or when it completes, the network seems to behave normally.

lspci says this about my network controller:

Intel Corporation PRO/Wireless 4965 AG or AGN Network Connection (rev 61)

[Martijn Faassen (faassen)]

Following the instructions in #9 here seems to fix my problems:

http://www.intellinuxwireless.org/bugzilla/show_bug.cgi?id=1623

[Martijn Faassen (faassen)]

Following the instructions in comment #9 here seems to fix my problems:

http://www.intellinuxwireless.org/bugzilla/show_bug.cgi?id=1623

[Gary Mansell (garymansell)]

I am getting this same issue with the released version of Hardy Heron,

I have wired network (BCM5755M) and transfers via nfs (v2 & v3), ftp, scp etc all hang after about 1 minute of transferring.

The system then goes to 99% wait state and becomes completely unresponsive.

This must be an important bug to fix?

Regards

Gary

[Gary Mansell (garymansell)]

Some more info - this might be related to x86_64 as I have just installed Hardy x86_64 and I did not have the issue with Gutsy i686

My Hardy kernel is - Linux grma-lap 2.6.24-16-generic #1 SMP Thu Apr 10 12:47:45 UTC 2008 x86_64 GNU/Linux

Rgds

Gary

[Messelink (messelink)]

After upgrading from Gutsy to Hardy (Desktop, 32bit) I'm having the same problems. I have no NFS or Wireless network and the problem only seems to occur with torrents active (and I have tried a lot of different torrent clients).

The pc is used as a local server (although with the desktop version of Ubuntu installed) and the system completely freezes, which can only resolved by a hard reset.

The on-board ethernet controller is used to connect to the outside world and the extra PCI ethernet card connects to the local LAN. It has an AMD Athlon 2400+ on an Abit AN7 mainboard and 512 MB RAM. I'm using kernel 2.6.24-16-generic and 1GiB swap.

I'm going to try and swap the two network cables and use the on-board network for the external link and the PCI card for the local network and see of this helps.

[Sean Corbett (seanbutnotheard)]

My machine has two things in common with your setup, hardware-wise: the AMD Athlon (mine is an MP system) and the on-board ethernet (not wireless). I'm also running the 2.6.24-16 (actually maybe 17? the latest Hardy) kernel. Oddly, torrents work fine for me, but any heavy local traffic (such as a multi-GiB FTP or SMB backup) cause a hard freeze.

My mainboard is a Tyan S2466, AMD-760 MPX chipset, on-board 3Com 3C920 ethernet. What ethernet controller(s) do you have? Can others report their hardware specs? Maybe we can isolate a common thread.

[Messelink (messelink)]

Just tried the -rt kernel image, without success. Computer froze in the middle of the night with a lot of torrents active. Can't find anything in the logs.

The onboad NIC is a 10/100 Fast Ethernet via RealTek RTL8201BL, the PCI card is an old 10/100 card, probably a 3com.

Oh, and I forgot to say that I am using a SATA disk in this system.

[Ivan Garcia (ivangarciapr)]

This also occurs to me. Running Hardy-32 with AMD 64 5200+, asus m2n-e motherboard, ralink wireless.

Freeze is total (ctrl-alt-backspace, f-keys, and Alt-SysRq REISUB do nothing). Does not produce any messages in system log.

[Ivan Garcia (ivangarciapr)]

Update: switching to ndiswrapper driver for my ralink (rt61) card appears to have done away with the issue for me.

[BongoMan (pascal-bongosoft)]

Same issue for me, running kubuntu hardy-32 on my desktop-PC. I can reproduce the freezes by doing high-speed downloads (> 5MB/s) for a while ( < 1 min ) either with wget, scp or ktorrent. The freeze is complete, I have to switch off the system with the power-button. Num-lock and scroll lock are blinking. Nothing suspicious is logged. I have a wired connection, no ndiswrapper.

Ethernet controller (onbard): Marvell Technology Group Ltd. 88E8001 Gigabit Ethernet Controller

uname -a: Linux bop7 2.6.24-17-generic #1 SMP Thu May 1 14:31:33 UTC 2008 i686 GNU/Linux

This is a very important bug, it needs to be fixed!!!

[BongoMan (pascal-bongosoft)]

sorry, it is Caps-lock and scroll-lock which are blinking, not num-lock... Curiously, I seem to have no freezes when downlowding with firefox.... Please ask if you need more informations about my system, I really would like to be helpful in fixing this bug :) Currently, I would not recommend to anyone to install ubuntu on a server with high netwok traffic...

[BongoMan (pascal-bongosoft)]

This could be helpful: I logged in in text mode without X/KDE and downloaded a big file with scp (with 10 MB/s). Again I had the hangup, but fortunately this time I got an error message on stout! Here are the last 3 lines:

Bad EIP value
EIP: [<002a8c29>] 0x2a8c29 SS: ESP 0068:c041de7c
Kernel panic - not syncing: Fatal excetion in interrupt

There is a longer Call Trace before, but I cannot copy/paste it, because the computer is frozen. Please ask if you need more infos.

I will try to reproduce the error this afternoon and redirect stderr/stdout to a file and send the output here.

[Chow Loong Jin (hyperair)]

BongoMan wrote:
> This could be helpful: I logged in in text mode without X/KDE and
> downloaded a big file with scp (with 10 MB/s). Again I had the hangup,
> but fortunately this time I got an error message on stout! Here are the
> last 3 lines:
>
> Bad EIP value
> EIP: [<002a8c29>] 0x2a8c29 SS: ESP 0068:c041de7c
> Kernel panic - not syncing: Fatal excetion in interrupt
>
> There is a longer Call Trace before, but I cannot copy/paste it, because
> the computer is frozen. Please ask if you need more infos.
>
> I will try to reproduce the error this afternoon and redirect
> stderr/stdout to a file and send the output here.
>

As far as I know, kernel panic outputs cannot be redirected to a file
because it doesn't come from the said application. However, you can poke
around the /var/log/dmesg.log and /var/log/syslog during your next boot up.

--
Hyperair

[BongoMan (pascal-bongosoft)]

yes you're right, i didn't manage to redirect this to a file. I am now burning the kubuntu DVD to check if the error also occurs when booting from the live-DVD. I will also try it on another PC to exclude some hardware issues... I'll report here soon...

[BongoMan (pascal-bongosoft)]

sorry for all my comments, finally it was a hardware problem... I had the freezes with the live DVD too, but only on this machine. I tried it on an another PC with identical hardware and there were no freezes. Now I could find a PCI-Network adapter, and using it instead of the onboard controller I don't have the freezes anymore. Sorry for having trashed this thread.

[Sean Corbett (seanbutnotheard)]

This bug seems to be Ubuntu-kernel (or Ubuntu-modules) specific however... I can reboot into System Rescue CD, which uses 2.6.24, and do very large rsync backups via smb and nfs without any trouble. I have seen some kernel updates appear in backports since I last tried, however, so I should update and try again.

[Todd Ross (brainsick)]

I was experiencing many of these same symptoms while running Ubuntu 8.04 on a Lenovo Thinkpad T60. Happened with both Transmission and Azureus. Hard locks. No messages in any log I could find. Finally, getting fed up, I tried installing openSUSE 11.0 and I'm happy to report that I'm no longer experiencing the problems. This definitely seems to be Ubuntu specific. It would be great if the Ubuntu team would acknowledge this problem and assign a resource to resolve it.

[PaNiek (kubuntobugs)]

I confirm the bug.

I can reproduce it by copying a directory of 8.5 GB from a mounted NFS drive to a local drive on my Compaq Pressario V6850ED. Or by downloading the directory with FTP. Even by copying it from the mounted NFS drive to the same mounted NFS drive(!). I can do so from 2 different servers (one AMD-64, one Pentium4). I can do it with Kubuntu 8.04 AMD64 as installed on the V6850ED. Or by running Kubuntu 8.04 I386 live from the CD. I use different physical locations on our network with different UTP cables. I tried many combinations (not all of them) and the result is always the same: At 3% to 55% in the copy process the V6850ED freezes with a blinking capslock light.

However, if I replace the V6850E by a Compaq Pressario C760ED I can copy the same directory from the same mounted NFS drive without any problem. I tested this with Kubuntu 8.04 I386 as well as with Kubuntu 8.04 AMD64 installed. The C760ED seems a tropical machine that never freezes.

The Compaq Pressario VG850ED has an AMD Turion 64 X2 TL60 processor @ 2GHz and 3072 MB DDR2 SDRAM.
The Compaq Pressario C760 has an Intel Pentium T2370 dual core processor @ 1.73 GHz and 2048 MB DDR2 SDRAM.

I hope this helps the Ubuntu team to resolve this problem.

[John Ward (automail)]

I want to back up what "BongoMan" said above, in relation to the Marvell Yukon 88E8001 causing problems. Ever since I started using Ubuntu (I started at Gutsy Gibbon 7.10, currently on Hardy) I have had kernel panics (a complete system lockup / crash / freeze and blinking Caps lock and Scroll lock lights) whenever I ran torrent programmes like Azureus or Transmission and less often when I ran Firefox 2 or 3 and YouTube. Throughout that time I used the Marvell Yukon 88E8001 LAN adapter. The torrent programmes would run for 10 minutes to 2 hours and then kernel panic, Transmission was occasionally better than Azureus at staying stable for longer but eventually the system would panic. Like I say also watching videos on YouTube was a common source of crashes and occasionally just normal browsing - i.e. internet activity.

Yesterday I remembered I had a second LAN port on my PC (an nVidia CK804 Gigabit adapter) and I decided to change over to it to see if things would improve and by gum they did. I have been able to run torrent (Azureus) straight for almost two days now without a hiccup. Unfortunately I can't say I am having the same problems in relation to picking up DHCP request or connection drops. My 88E8001 has never dropped once or failed to fulfill a DHCP request (i.e. receive a DHCP I.P.). Here is my hardware:

- ASUS A8N-SLI Deluxe Motherboard
      - Marvell Yukon 88E8001 Gigabit LAN adapter (onboard)
      - nVidia CK804 Gigabit LAN adapter (onboard)
      - nForce4 chipset

- AMD Athlon 3700+ San Diego

- 1GB Corsair XMS4400 DDR RAM

- nVidia 7800GTX PCI-E graphics card

[David V (verstraetendavid)]

I can confirm that this problem is still occuring on my Kubuntu 8.04 AMD64 machine.

Using a ralink rt61 (not using ndiswrapper).
Linux lx-p45 2.6.24-19-generic #1 SMP Wed Jun 18 14:15:37 UTC 2008 x86_64 GNU/Linux
filesystem ext3

The problem occurs almost immediatly after I start the downloading of a large file using adept or any torrent client causing the system to hang. I can't find any traces of this in any log, but if they are needed let me know so I can send them.

[Wolfgang Glas (wglas)]

As stated by me in bug #204996 under

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/204996/comments/256

there are some deadlock fixes in 2.6.25, which likely apply to this problem. Unfortunately, no kernel developer steped up so far to comment on the patch mentioned in my comment.

Hopefully, we get some feedback in the next few days.

   Regards, Wolfgang

[Erik Meitner (e.meitner)]

I too have this problem with 8.04 on a Thinkpad T60(8744-5BU). It started about two to four months ago. I wish I remembered exactly when.
I did not do any kind of bug report then as I had to take the time to eliminate the possibility that it was due to the bleeding-edge drivers I generally use(madwifi and radeonhd).

The symptoms of the lockup are:

* NO blinking caps light
* a total system freeze,
* unresponsive to sysrq
* If the hard disk was active at the time the disk light remains lit
* nothing in the logs

I tried using the distribution's radeon and fglrx x.org drivers, I completely disabled the madwifi drivers, and I tried running older Hardy beta kernels I had installed such as 2.6.24-12-generic but still experienced lockups.

Then I noticed that the lockups seem to occur while the laptop is being backed up. I use BackupPC(rsync as the transfer protocol) to backup the laptop both at home and at work. The lockups happen 80% of the time during a backup at home and not at all at work. I disabled backups at home and it seemed to help. There were a few lockups that did happen so this may indicate it is a certain pattern of network activity rather than an application that causes it.

Then I did some Bittorrent downloads using gnome-btdownload and had lockups quite consitently with that too.(I don't use Bittorrent at work so I can't really test to see if there is a network-specific issue here.) I then did some Bittorrent downloads using btdownload-curses from the console and not switching away from the console. No lockups. BUT, I found that after doing this(with a full or partial BT download) and switching back to the Gnome desktop the system would lock after a few minutes. Also interesting is that I recently did a long Bittorrent(gnome-btdownload) download from a slow seeder(1-5 KB/s. I know that is REAL slow, but it hardly matters when I leave the laptop on ovrenight) and did not have a lockup UNTIL the system experienced high network utilization.(MythTv I think)

I have had only two lockups while running MythTv(viewing HD video) with its high network utilization.(one mentioned above) There have been a few lockups during regular web browsing and network utilization was low. Again, it does not always appear to be a matter of the amount of network utilization, but some other factor.

I must also note that I am ALWAYS on the WIRED ethernet when these lockups happen. I wonder if there is an e1000 driver issue? Or an interaction between e1000 and some junk switches?(I've got a no-name 10/100 switch)

I have not been able to obtain any metrics that would indicate a problem. I was monitoring the values in /proc/slabinfo and did see the number of active objects for "dentry" increase quite a bit during Bittorrent downloads and backups. But this is where everything is black magic to me. I can get you these numbers if you thinkg it will help.

Running kernel 2.6.24-19-generic.

Bug 214108 may be a candidate for the duplicate bugs list.

[David V (verstraetendavid)]

I've tried using ubuntu 8.04 32bit and the problem is exactly the same there. Total system hang when downloading large files.

I've now installed an ubuntu 7.04 and the exact problem doesn't seem to reproduce itself.
I've succesfully downloaded a large number of updates(about 400MB if i have to guess, didn't check how much exactly) at 493KBps without hang under the following kernel:
Linux lx-ub710 2.6.22-14-generic #1 SMP Sun Oct 14 23:05:12 GMT 2007 i686 GNU/Linux
I made sure not to reboot to give me some change to test if downloading a torrent would not reproduce the problem. And although my network connection stops functioning properly with a message to submit a bug to rt2x00.serialmonkey, the pc doesn't hang in itself, which was by far the most troubling aspect.

I hope this is another problem I'm encountering now, one that can be solved. Although I would love to help fix this problem, the fact that it seems to be ongoing since november 2007 doesn't seem promising and I need this new machine to be productive.

[jchronakis (jchronakis)]

I can confirm the same or similar problem.

When copying large files over the network (ethernet, cifs mounted share) after a few seconds to a few minutes the system hangs with kernel panic (blinking keyboard lights). This happens all the time and I never managed to copy more than a couple of GBs.

I did a copy from the console and managed to get a picture of the panic. Unfortunately shift-pgup was not working so it is not the whole dump.

My system is: 2.6.24-19-generic #1 SMP Fri Jul 11 23:41:49 UTC 2008 i686 GNU/Linux

Hardware:
M/B: ASUS P4P800-E Deluxe
LAN: Marvell 88E8001 Gigabit Ethernet controler
WiFi: Belkin G Wireless PCI Adaptor

While this happens only with ethernet traffic, the kernel panic seems to be related with the wireless:
EIP: [<f8bd0c52>] ieee80211_tx_status_irqsafe+0x12/0x120 [mac80211] SS:ESP 0068:dd36ff14

[jchronakis (jchronakis)]

I think that the problem has been resolved in latest kernel versions.

I installed the hardy modules backports (linux-backports-modules-hardy) and since then I have attempted to copy twice about 10Gigs from a samba share without any problem.

[John Ward (automail)]

So in other words you did a "sudo apt-get install linux-backports-modules-hardy" or do you mean recompiling? I'd like to try it because now Ubuntu is crashing for any reason at the moment, not just network related - run RhythymBox for a while and do something like press "Next Track" and kernel panic! Possibly Firefox is running in the background.

[jchronakis (jchronakis)]

John,

I didn't compiled anything. I just used synaptic to install the linux-backports-modules-hardy package and then rebooted.

I hope it helps

[John Ward (automail)]

Cheers Jchronakis,

I think thats what I meant when I said doing "sudo apt-get install linux-backports-modules-hardy" but of course you can do it through synaptic package manager also. Thanks for the suggestion, I'll try it now even.

[John Ward (automail)]

Unfortunately no difference for me, still locking up. I think this bug here is very similar: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/204996

[frenzie (frenzie)]

Installing the backports worked for me!

[frenzie (frenzie)]

hold on, i spoke too soon.
Still having the same trouble, copying files over samba, freezes all the time
evenwhen i try and copy to the ubuntu machne it freezes

[Erik Meitner (e.meitner)]

Adding to my comment above from 2008-07-15:
Another situation that consitently causes this is RDP protocol(Windows remote desktop) over IPSEC using the wired ethernet(e1000). Does NOT happen over wireless.

What a few of us need to do is get this working on our systems and provide more data for the package maintainers to work with:
https://blueprints.launchpad.net/ubuntu/+spec/linux-kernel-crash-dump

[Erik Meitner (e.meitner)]

From my testing I have determined that the lockup kills the kernel. The "softdog" watchdog fails to reboot the system system and sysrq commands do not seem to be able reboot either. Thus I think that kdump is of no use in this situation. I'll try to see if there is some serial console output to capture. I am able create the lockups quite consistently as stated above.(RDP over IPSEC using e1000 ethernet)

[Erik Meitner (e.meitner)]

My suspicion of an interaction between the e1000 ethernet and a junk switch(see my comment on 2008-07-15) may have been correct. I replaced the old switch with a new higher-quality one and cannot reproduce the problems. In my opinion this is still a bug that must be addressed as we cannot have Ubuntu crashing just because some network hardware is junk.

[Wolfgang Glas (wglas)]

As stated in https://bugs.launchpad.net/ubuntu/+source/linux/+bug/204996/comments/256 I also came to the conclusion that this is a deadlock inside the TCP/IP stack in some corner cases of the state machine. (In my case triggered by a lousy UMTS dialup connection...)

I even identified a patch inside 2.6.25, which might sure the problem, however I never got an answer from a kernel developer on this posting :(

[David V (verstraetendavid)]

Erik, define higher quality please? The machine I have that encountered this problem was connected to a 150 euro (200+ dollar) router, and the other machine I have which also runs Ubuntu 8.04 has never had this problem although it's connected to the same appliance. And I use that one as my primary download machine, seeing as it also functions my fileserver at home.

[Demothesis (demothesis)]

I've been experiencing this same issue on Ubuntu Ibex beta i386, up to date as of this morning (10/11/2008). The problem seems to occur more frequently when downloading torrents. Doesn't matter if I'm using Vuze or Transmission. It also occurs when transferring files over the network via Samba. Logs show normal operation and then the restart after it hard locks.
When it does freeze I have no response from the keyboard at all, and caps lock and scroll lock blink as reported by a few others in this thread. The length of time it takes to freeze seems to vary from ~20 minutes up to a full 24 hours.
Hardware is a Dell Vostro 1500 with an Intel 4965 wireless card. Occurs using multiple routers/access points.

[Gary Mansell (garymansell)]

It's criminal that this is still a bug in the latest release of Ubuntu Ibex, this has been a known issue for at least a year now. How can any Operating System than freezes/hangs on a file copy be considered a usable OS?

This bug prevents me using Ubuntu on a whole office worth of machines.

----- Original Message ----
From: Demothesis <email address hidden>
To: <email address hidden>
Sent: Saturday, 11 October, 2008 19:36:55
Subject: [Bug 147464] Re: Heavy network activity (eg: torrent/nfs file transfers) causes Hard System Locks and/or Network Freezes.

I've been experiencing this same issue on Ubuntu Ibex beta i386, up to date as of this morning (10/11/2008). The problem seems to occur more frequently when downloading torrents. Doesn't matter if I'm using Vuze or Transmission. It also occurs when transferring files over the network via Samba. Logs show normal operation and then the restart after it hard locks.
When it does freeze I have no response from the keyboard at all, and caps lock and scroll lock blink as reported by a few others in this thread. The length of time it takes to freeze seems to vary from ~20 minutes up to a full 24 hours.
Hardware is a Dell Vostro 1500 with an Intel 4965 wireless card. Occurs using multiple routers/access points.

--
Heavy network activity (eg: torrent/nfs file transfers) causes Hard System Locks and/or Network Freezes.
https://bugs.launchpad.net/bugs/147464
You received this bug notification because you are a direct subscriber
of the bug.

[Demothesis (demothesis)]

I'm not entirely certain about anyone else. But I haven't had this problem in any other release since 7.04, and don't notice it using an intel 3945 wireless G card in Intrepid beta. It almost makes me suspect that my issue is something to do with wireless N support in the new kernel, but I don't know enough to actually investigate it.

[Chow Loong Jin (hyperair)]

On Mon, 2008-10-13 at 07:18 +0000, Gary Mansell wrote:
> It's criminal that this is still a bug in the latest release of Ubuntu
> Ibex, this has been a known issue for at least a year now. How can any
> Operating System than freezes/hangs on a file copy be considered a
> usable OS?
>
> This bug prevents me using Ubuntu on a whole office worth of machines.

>
Until someone can identify exactly which machines have the problem and a
common factor between the machines in terms of hardware, it will be
close to impossible to fix this bug. Instead of complaining, why don't
you take the initiative to do just that?

None of my machines face this bug.
--
Chow Loong Jin

[Chow Loong Jin (hyperair)]

Marking this bug on nfs-utils as invalid because it seems to not be dependent on nfs, but on network transfers in general.

[Wolfgang Glas (wglas)]

As I stated earlier, this bug is related to TCP/IP packet drops, which may be triggered by several factors such as:

- A dialup connection through GPRS/UMTS
- A switch with a primitive drop-tail policy under heavy load.
- Very, very many simultaneous TCP/IP connections.

I already pointed out an applicable patch from the 2.6.25 release series. However, I never got any response, whether my assumption may be valid or not. We are observing this deadlock on various machines, the only common denominator so far seems to be that we are running a x86_64 version of ubuntu hardy. I never observed a dead-lock under i386, but this may be due to the fact that we have only 20% of our hardware running under i386.

  Regards,

     Wolfgang

[Gary Mansell (garymansell)]

I am not convinced - I see this on a x86_64 hardy install just copying files across the network and we have a 3com network environment that is pretty industry standard

----- Original Message ----
From: Wolfgang Glas <email address hidden>
To: <email address hidden>
Sent: Monday, 13 October, 2008 13:06:46
Subject: [Bug 147464] Re: Heavy network activity (eg: torrent/nfs file transfers) causes Hard System Locks and/or Network Freezes.

As I stated earlier, this bug is related to TCP/IP packet drops, which
may be triggered by several factors such as:

- A dialup connection through GPRS/UMTS
- A switch with a primitive drop-tail policy under heavy load.
- Very, very many simultaneous TCP/IP connections.

I already pointed out an applicable patch from the 2.6.25 release
series. However, I never got any response, whether my assumption may be
valid or not. We are observing this deadlock on various machines, the
only common denominator so far seems to be that we are running a x86_64
version of ubuntu hardy. I never observed a dead-lock under i386, but
this may be due to the fact that we have only 20% of our hardware
running under i386.

  Regards,

     Wolfgang

--
Heavy network activity (eg: torrent/nfs file transfers) causes Hard System Locks and/or Network Freezes.
https://bugs.launchpad.net/bugs/147464
You received this bug notification because you are a direct subscriber
of the bug.

[effell (effell)]

I encountered this bug on my (non x64) Thinkpad using a 3G/GPRS modem. As I stated then, "Updating the kernel from 2.6.22-14-generic to 2.6.23.9 seems to have eliminated this bug from my system."

The solution (to at least one bug) is in the diff between those Ubuntu-generic and the kernel.org kernels.

I have since stopped using said modem, so I can't help with further testing.

[Johan Kröckel (jnkl)]

I observed this bug on a thinkpad x60s with a core duo (i386) and intel nics.

[ugtar (uokrent)]

I have only experienced this bug when connecting using 802.11n. Otoh,
my system doesn't completely lock up, just the network interface stops
responding. The bug I am describing does sound a bit different from
this one...
I'm using a broadcom wireless adapter with the latest (1.53?)
ndiswrapper and a driver that works on windoze. I believe what I'm
experiencing is a kernel bug though because once the network interface
goes down nothing short of a reboot will get it working again. I guess
I should mention that I'm on 32bit hoary, but I've had the same exact
experience with 64bit, and on both fiesty and gutsy on this particular
machine, and using several drivers and several versions of
ndiswrapper. My workaround is to just only connect using 802.11g or b
and everything works just peachy (if not a little slower than I'd like).

On Oct 13, 2008, at 12:44 AM, hyperair <email address hidden> wrote:

> On Mon, 2008-10-13 at 07:18 +0000, Gary Mansell wrote:
>> It's criminal that this is still a bug in the latest release of
>> Ubuntu
>> Ibex, this has been a known issue for at least a year now. How can
>> any
>> Operating System than freezes/hangs on a file copy be considered a
>> usable OS?
>>
>> This bug prevents me using Ubuntu on a whole office worth of
>> machines.
>
>>
> Until someone can identify exactly which machines have the problem
> and a
> common factor between the machines in terms of hardware, it will be
> close to impossible to fix this bug. Instead of complaining, why don't
> you take the initiative to do just that?
>
> None of my machines face this bug.
> --
> Chow Loong Jin
>
> --
> Heavy network activity (eg: torrent/nfs file transfers) causes Hard
> System Locks and/or Network Freezes.
> https://bugs.launchpad.net/bugs/147464
> You received this bug notification because you are a direct subscriber
> of the bug.

[Johan Kröckel (jnkl)]

I suffered from this bug under both nics (e1000 and iwl3945), on both reproducable, sometimes it takes 5 mins, sometimes 8 hours but I didnt download one bigger file (>1G) over bittorrent without a freeze.
It happened sometimes that bittorrent ran for some hours, without an user, and the freeze happened just when I startet firefox and began surfing.
But I stopped downloading using bittorrent for some weeks and no freeze happened for that time, and I produced some heavy network traffic (copied 150GB with rsync over 100mbit ethernet more than once).

[Gary Mansell (garymansell)]

Looks like this may have got fixed for me in the Ubuntu x86_64 Intrepid Ibex release - I have just copied 4 GB from a server using nfs to my laptop and I have not had a hang. Pretty sure that I would have normally.

Fingers crossed....

Gary

[Annoyed Ape (jarkoian)]

Um, no... the problem persists. My machine is a single AMD Athlon Processor (32 bit of course!). It was happening with Hardy Heron, and before that with Gutsy Gibbon, and now of course with yesterday's upgrade to Intrepid Ibex. The kernel I'm running is (from uname -a):

Linux MythTVServer 2.6.27-7-generic #1 SMP Tue Nov 4 19:33:20 UTC 2008 i686 GNU/Linux

My machine drops the network connection to the world when transferring large files, I can make it happen in 2 or 3 seconds by trying to copy a large number of mp3s from a Windows share to the Ubunutu server through Samba. It can also happen, as a user stated above, by lots of network connections of little bandwidth when using MythTV. (This failure's occurance is unpredictable, and the server can last for weeks with no problems before the remote frontend complains about not seeing the backend anymore.)

I am using quality Netgear pro Gigabit switches. (The nice metal ones.) I've swapped network cards (Linksys PCI Gigabit cards) and used the onboard 100 MB VIA card as well. None of this eliminates the bug.

I've also sucessfully used the same network card on a x86_64 version of Ubuntu on another server with no problems! I've also used another x86_64 server with a built in network card with no problems.

The only things that are specific to my lock up is the motherboard/CPU... perhaps could it be the PCI bus controller? What about the BIOS's ACPI control? The machine has 6 hard drives with one PCI IDE soft-RAID controller, the onboard IDE controller, and 2 Hauppauge PVR350 PCI video capture cards. The results of "cat /proc/interrupts" shows the eth1 (Linksys Gigabit) card is sharing Interrupt 16 with one of the ivtv devices... but both work. There are no errors in the ring buffer (/bin/dmesg) or the syslog.

As this makes me want to hit my computer with a BFH, let me know what I can do to help track this down.

[Gary Mansell (garymansell)]

Well,

It has been a few days now and I have tried copying large files by NFS (v2 and v3 work now but did not before) and FTP and have not experienced the problems that have been bugging me all through the Hardy release. Nothing else has changed apart from the OS upgrade. The details of my SW/HW are:

Ubuntu Intrepid Ibex x86_64
High Quality 3Com Netowork infrastructure
Dell Latitude D630:
grma@grma-lap:~$ lspci
     00:00.0 Host bridge: Intel Corporation Mobile PM965/GM965/GL960 Memory Controller Hub (rev 0c)
     00:01.0 PCI bridge: Intel Corporation Mobile PM965/GM965/GL960 PCI Express Root Port (rev 0c)
     00:1a.0 USB Controller: Intel Corporation 82801H (ICH8 Family) USB UHCI Controller #4 (rev 02)
     00:1a.1 USB Controller: Intel Corporation 82801H (ICH8 Family) USB UHCI Controller #5 (rev 02)
     00:1a.7 USB Controller: Intel Corporation 82801H (ICH8 Family) USB2 EHCI Controller #2 (rev 02)
     00:1b.0 Audio device: Intel Corporation 82801H (ICH8 Family) HD Audio Controller (rev 02)
     00:1c.0 PCI bridge: Intel Corporation 82801H (ICH8 Family) PCI Express Port 1 (rev 02)
     00:1c.1 PCI bridge: Intel Corporation 82801H (ICH8 Family) PCI Express Port 2 (rev 02)
     00:1c.5 PCI bridge: Intel Corporation 82801H (ICH8 Family) PCI Express Port 6 (rev 02)
     00:1d.0 USB Controller: Intel Corporation 82801H (ICH8 Family) USB UHCI Controller #1 (rev 02)
     00:1d.1 USB Controller: Intel Corporation 82801H (ICH8 Family) USB UHCI Controller #2 (rev 02)
     00:1d.2 USB Controller: Intel Corporation 82801H (ICH8 Family) USB UHCI Controller #3 (rev 02)
     00:1d.7 USB Controller: Intel Corporation 82801H (ICH8 Family) USB2 EHCI Controller #1 (rev 02)
     00:1e.0 PCI bridge: Intel Corporation 82801 Mobile PCI Bridge (rev f2)
     00:1f.0 ISA bridge: Intel Corporation 82801HEM (ICH8M) LPC Interface Controller (rev 02)
     00:1f.1 IDE interface: Intel Corporation 82801HBM/HEM (ICH8M/ICH8M-E) IDE Controller (rev 02)
     00:1f.2 IDE interface: Intel Corporation 82801HBM/HEM (ICH8M/ICH8M-E) SATA IDE Controller (rev 02)
     00:1f.3 SMBus: Intel Corporation 82801H (ICH8 Family) SMBus Controller (rev 02)
     01:00.0 VGA compatible controller: nVidia Corporation Quadro NVS 135M (rev a1)
     03:01.0 CardBus bridge: O2 Micro, Inc. Cardbus bridge (rev 21)
     03:01.4 FireWire (IEEE 1394): O2 Micro, Inc. Firewire (IEEE 1394) (rev 02)
     09:00.0 Ethernet controller: Broadcom Corporation NetXtreme BCM5755M Gigabit Ethernet PCI Express (rev 02)
     0c:00.0 Network controller: Intel Corporation PRO/Wireless 3945ABG [Golan] Network Connection (rev 02)

Hopefully that info may help someone who knows about these things.

Anyway good luck to the rest of you that someone can get this fixed for you cos I know what a nightmare pain the bug is - Even Windows Vista can copy files over the network (I know its slow but it works) :-(

Rgds

Gary

[Tank (tank85)]

Same issue for me, T60 freezes at WLAN heavy load

More info:
Ubuntu release: Intrepid
Network controller: Intel Corporation PRO/Wireless 3945ABG [Golan] Network Connection
Kernel driver in use: iwl3945
Kernel: 2.6.27-7-generic

[Annoyed Ape (jarkoian)]

Here are my system files that may help in debugging. They include the results from the following commands:

uname -a
sudo lspci -vvv
cat /proc/interrupts
cat /proc/iomem
cat /proc/ioports
sudo lsmod

[Gary Mansell (garymansell)]

Just to prevent any confusion here, please note that my problems were not with wireless networking but with wired networking.

I was seeing the problem in Hardy x86_64 using wired ethernet (Broadcom Corporation NetXtreme BCM5755M) and with what looks like to me the tg3 driver.

It now appears to be resolved in Intrepid Ibex x86_64

Rgds

[Annoyed Ape (jarkoian)]

My connection is also a WIRED connection. To further clarify, I DO NOT experience a system lock-up, NOR do I see the blinking caps lock, scroll lock, num lock lights. What I experience is total network absence; as if you'd unplugged the machine from the switch.

If I do a "sudo ifconfig eth1 down" and then a "sudo ifconfig eth1 up" from the terminal, the network comes back to life.
A simple "sudo route add default gw 192.168.0.1" will allow DNS lookups to work again.

Pinging anything when the network has dropped shows a series of "network unreachable" messages.

[Johan Kröckel (jnkl)]

Annoyed Ape:
You didnt encounter the bug thats discussed here...
I suggest you open a new bug report.

[Annoyed Ape (jarkoian)]

Johan, thanks, I posted a new bug #296162. Hopefully the admins will agree with you!

[headlessspider (headlessspider)]

i experience this often when transferring large files to a nfs server. i am using an hp compaq nx6120 with 768 meg ram on a wired connection. i am using iterpid. this is a fresh install of 8.04 lts then upgraded to iterpid. when transfers stop, access to the network and the internet also stops. its as if the network connection has been unplugged. unlike the others i do not get any blinking lights. gnome is still responsive but when i try restarting the laptop it hangs somewhere in the middle of the shutdown/restart process and the mute button lights up. i would have to do a hard shutdown of the laptop. this happens approximate 67% of the time (when transferring large files).

this did not happen to me when i was using 8.04.

[Tony (tonybaca)]

Just had, what I think is the same problem. Running X86-64 8.10 Intrepid fresh insall. I mounted an NFS volume from another system (SUSE) via wired connection. Tried to copy over 300G video files. Each is about 1G. System transfered about 30% then hard lock, TOP was still runing. If I stop TOP, then I can't execute any other commands. I tried to SSH but that does not work. I am forced to power cycle. Tried this severarl times and each time it hard locks the client computer. The attached file is something I found in the log

[hajkos (hajkos)]

Submitting the same problem. Intrepid amd64 crashes under heavy load & more network connections (samba, ssh, headless virtualbox). Used to be Hardy, but then I performed a dist-upgrade, because the bug occured at Hardy too. Thought it would help :( Sometimes I can push 50 gigs over samba straight, sometimes it crashes after 100 megs. It's an ASUS-M2A-MX board with integrated Attansic (atl1) card and an Intel 1000 (e1000) card. Detailed info enclosed in the attached file.
Hope this may help.

The /var/log/messages shows the following:

Nov 30 12:08:14 homer kernel: [ 65.393126] e1000: eth1: e1000_watchdog: NIC Link is Up 1000 Mbps Full Duplex, Flow Cont
rol: RX/TX
Nov 30 12:08:14 homer kernel: [ 65.801001] NET: Registered protocol family 10
Nov 30 12:08:14 homer kernel: [ 65.801356] lo: Disabled Privacy Extensions
Nov 30 12:08:15 homer kernel: [ 71.976814] iSCSI Enterprise Target Software - version 0.4.15
Nov 30 12:08:15 homer kernel: [ 71.976892] iscsi_trgt: Registered io type fileio
Nov 30 12:08:15 homer kernel: [ 71.976894] iscsi_trgt: Registered io type blockio
Nov 30 12:08:15 homer kernel: [ 71.976896] iscsi_trgt: Registered io type nullio
Nov 30 12:10:08 homer kernel: [ 184.698833] vboxdrv: fAsync=1 offMin=0xfbbf3 offMax=0xfbbf3
Nov 30 12:10:08 homer kernel: [ 184.698842] vboxdrv: TSC mode is 'asynchronous', kernel timer mode is 'normal'.
Nov 30 12:28:14 homer -- MARK --
Nov 30 12:35:20 homer kernel: [ 1697.047794] md: md3: resync done.
Nov 30 12:35:20 homer kernel: [ 1697.203374] RAID1 conf printout:
Nov 30 12:35:20 homer kernel: [ 1697.203379] --- wd:2 rd:2
Nov 30 12:35:20 homer kernel: [ 1697.203381] disk 0, wo:0, o:1, dev:sda6
Nov 30 12:35:20 homer kernel: [ 1697.203383] disk 1, wo:0, o:1, dev:sdb6
Nov 30 12:38:46 homer kernel: [ 1903.073282] program smartctl is using a deprecated SCSI ioctl, please convert it to SG_I
O
Nov 30 12:38:46 homer kernel: [ 1903.074556] program smartctl is using a deprecated SCSI ioctl, please convert it to SG_I
O
Nov 30 12:40:04 homer kernel: [ 1980.761244] kjournald starting. Commit interval 5 seconds
Nov 30 12:40:04 homer kernel: [ 1980.762724] EXT3 FS on sdd1, internal journal
Nov 30 12:40:04 homer kernel: [ 1980.762731] EXT3-fs: recovery complete.
Nov 30 12:40:04 homer kernel: [ 1980.762734] EXT3-fs: mounted filesystem with ordered data mode.

>>>>> STARTED COPYING OVER SAMBA AND AFTER A FEW MINUTES CRASH OCCURED HERE <<<<<<<<

Nov 30 13:08:13 homer syslogd 1.5.0#2ubuntu6: restart.
Nov 30 13:08:13 homer kernel: Inspecting /boot/System.map-2.6.27-9-server
Nov 30 13:08:14 homer kernel: Cannot find map file.
Nov 30 13:08:14 homer kernel: Loaded 49686 symbols from 66 modules.
Nov 30 13:08:14 homer kernel: [ 0.000000] Initializing cgroup subsys cpuset
Nov 30 13:08:14 homer kernel: [ 0.000000] Initializing cgroup subsys cpu
Nov 30 13:08:14 homer kernel: [ 0.000000] Linux version 2.6.27-9-server (buildd@yellow) (gcc version 4.3.2 (Ubuntu 4.3.2-1ubuntu11) ) #1 SMP Thu Nov 20 22:56:07 UTC 2008 (Ubuntu 2.6.27-9.19-server)
Nov 30 13:08:14 homer kernel: [ 0.000000] Command line: root=/dev/md2 ro quiet splash

[Kunin (kunin)]

I have had this same issue since 7.10, currently on 8.04 and after almost a year finally figured out that it always seems to hard lock when I've got a ton of torrent activity. The up/down speed it locks on is always different, so I'm thinking it has to do with the number of connections present at the time and not the about of bandwidth being used.

To be sure I have tested every piece of hardware, replacing my NIC (no effect, except I upgraded so now I get even better speeds) and replaced my PSU due to a slight fluctuation in one of the 5v rails (within limits, but barely) which also had no effect. CPU, heat, RAM, HDs, etc all tested fine. So this is definitely a Linux/Ubuntu issue.

My issues are exactly as above, seemingly random hard locks requiring the use of the reset/power button. I cannot confirm cases of num lock/caps lock/etc lights blinking as my wireless keyboard doesn't have those lights.

Please fix this issue, it should be of a relatively high priority due to the nature of the issue and the possible public (read non-Linux users whom wish to try Ubuntu) assumptions that Ubuntu is not a stable OS. I personally have fiber at home, and up until now (when I realized where the issue is) I would happily leave any and all torrents running until these was not a single peer left, having a constant 3-5Mbps (and higher) upload going. I've tried other torrent clients (I prefer Azureus) with the same resulting hard locks.

[effell (effell)]

This is a serious bug. I experienced it with the generic kernel a good while ago
and managed to get around it by compiling my own kernel from kernel.org.

root@localhost:~$ uname -a
Linux localhost 2.6.23.9 #1 PREEMPT Fri Dec 7 07:16:40 WET 2007 i686 GNU/Linux

So, I'm still using Ubuntu 7.10 with this kernel and maybe I'm missing all kinds of
security upgrades. Not good.

PS This other kernel also solves bug #43092, which is also alive and kicking.

[Gary Mansell (garymansell)]

I think I spoke too early when I said that this was fixed in Intrepid x86_64....

I think I am seeing this again. I am downloading a file from an ftp site with Firefox and my system keeps freezing for approx 5-10 secs at a time every couple of minutes. The Windows all go dark and then come back to alive again which is what I was seeing before.

The download is only running at 100KB/s sois only using minimal bandwidth on my corporate 30MB/s Internet connection.. My laptop is plugged into a 1GB network port at the core of our 3Com network.

I declare this still a problem - but the occurrence is different to what I was suffering before: it is a lot less and the system seems to be able to un-stick itself rather than requiring a hard reset.

Still Bl**dy frustrating though - my system has locked up about 10 times whilst writing this post. Almost unusable!!

Gary

[Qiwichupa (scorpion-matrixagents)]

I can confirm that the same bug on my Dell Inspiron 1300 with Kubuntu 8.04 (lastest updates included). System going down from the coping files, or watching the movies (with quickly pause/play cycle) from share.

[xteejx (xteejx)]

In order to get this to a Triaged state, the Kernel Team will need the following information.

Qiwichupa, or anyone else in Hardy: Can you attach these following as separate attachments please:

uname -a > uname-a.log
cat /proc/version_signature > version.log
dmesg > dmesg.log
sudo lspci -vvnn > lspci-vvnn.log

Note, if this is a problem in Jaunty you can use the following command:

apport-collect -p linux-image-`uname -r` 147464

Thank you.

[headlessspider (headlessspider)]

Architecture: i386
DistroRelease: Ubuntu 9.04
HibernationDevice: RESUME=UUID=033dee7e-6241-4b76-8c46-194bd632bcaf
MachineType: Hewlett-Packard HP Compaq nx6120 (PV160PA#UUF)
Package: linux-image-2.6.28-11-generic 2.6.28-11.42
PackageArchitecture: i386
ProcCmdLine: root=UUID=8395a20c-0276-4ba5-a2af-a1dee65495f9 ro quiet splash
ProcEnviron:
 SHELL=/bin/bash
 LANG=en_PH.UTF-8
ProcVersionSignature: Ubuntu 2.6.28-11.42-generic
Uname: Linux 2.6.28-11-generic i686
UserGroups: adm admin audio cdrom dialout dip floppy fuse lpadmin plugdev video

[headlessspider (headlessspider)]

ey teej,

i have done what you requested (am running jaunty)

just an update. i was copying approximately 220 megabytes of files to a samba server via sshfs and it just stopped at the point where 32.4 megabytes have been transferred. i left it alone for about 20 minutes and when i got back it was still there.

i'm trying to confirm if it also happens on an hp compaq nc6320 also running on jaunty.

[xteejx (xteejx)]

headlessspider, thank you very much for updating us, very much appreciated. I shall mark this as Triaged and set High importance, this will undoubtedly affect a lot of users, especially if this has been a problem since Feisty.

[headlessspider (headlessspider)]

i can confirm that this bug appears also in an hp compaq nc6320 running jaunty. so far, the only thing that's the same on an nc6320 and the nx6120 that i'm using is the network driver which is labeled tg3.

using kernel 2.6.28-15-generic on both laptops

-- noel

[headlessspider (headlessspider)]

one thing i did notice lately is that the copy would stop (not hang) for about 2 minutes, the system would say that eth1 is disconnected then the system prompty reconnects and then the copy proceeds. should the copy stop again the system then disconnects and then reconnects again until the copy is finished.

yes, the file gets copied eventually but everyone here knows that eth1 shouldn't reset just to have the file copy continue and finish.

[I Kovalev (iakovalev)]

Have being running Ubuntu (9.04 to 9.10) on Toshiba Satellite L40-139 laptop for about a year. Heavy network traffic systematically triggers a crash (either kernel panic or segmentation fault). Among applications that have triggered system crash are: firefox (all versions since 3.2.3), system update manager, a Java-based application downloading via http/ftp as well as bittorrent client and wget. Usually the problem appears within 20-30 min while downloading at speed of 150kBytes/s or more. Just before the moment of crash, tons of messages like:
Feb 3 02:30:04 kovalev-home kernel: [35474.511460] recvmsg bug: copied BFA5DAF9 seq BFA5E0A1
fill up system logs on pair with application-related warning. Sometimes (fifty-fifty) there is no immediate crash after warnings have appeared, but part of downloaded data have become corrupt.

The problem has been observed first with kernel 2.6.28-13 but still persists through current 2.6.31-19, all have been installed from official Ubuntu repositories. Network connection is via Atheros AR5007eg wireless (reported as AR5001 by ath5k). I've experienced this issue with several different routers and providers, though have never got a chance to try a wired connection. As Windows Vista had never crashed by similar network traffic on this laptop, the issue doesn't look like a hardware fault...

An example of warning message right before crash (triggered by bittorrent client):
---------------
Feb 2 12:51:38 kovalev-home kernel: [76478.454177] WARNING: at /build/buildd/linux-2.6.31/net/ipv4/tcp.c:1408 tcp_recvmsg+0xa49/0xb20()
Feb 2 12:51:38 kovalev-home kernel: [76478.454179] Hardware name: Satellite L40
Feb 2 12:51:38 kovalev-home kernel: [76478.454181] Modules linked in: sbp2 xt_tcpudp aes_i586 aes_generic binfmt_misc ppdev nls_iso8859_1 nls_cp437 vfat fat joydev snd_hda_codec_analog arc4 ecb pcmcia snd_hda_intel snd_hda_codec snd_hwdep snd_pcm_oss snd_mixer_oss snd_pcm snd_seq_dummy iptable_filter snd_seq_oss ip_tables x_tables snd_seq_midi snd_rawmidi snd_seq_midi_event snd_seq snd_timer snd_seq_device psmouse serio_raw yenta_socket rsrc_nonstatic pcmcia_core snd shpchp ath5k mac80211 ath cfg80211 soundcore snd_page_alloc asus_laptop led_class video1394 raw1394 lp parport usbhid fbcon tileblit font bitblit softcursor 8139too 8139cp mii ohci1394 ieee1394 i915 drm i2c_algo_bit intel_agp agpgart video output
Feb 2 12:51:38 kovalev-home kernel: [76478.454230] Pid: 2357, comm: transmission Tainted: G W 2.6.31-18-generic #55-Ubuntu
Feb 2 12:51:38 kovalev-home kernel: [76478.454233] Call Trace:
Feb 2 12:51:38 kovalev-home kernel: [76478.454237] [<c014513d>] warn_slowpath_common+0x6d/0xa0
Feb 2 12:51:38 kovalev-home kernel: [76478.454242] [<c04d2439>] ? tcp_recvmsg+0xa49/0xb20
Feb 2 12:51:38 kovalev-home kernel: [76478.454245] [<c04d2439>] ? tcp_recvmsg+0xa49/0xb20
Feb 2 12:51:38 kovalev-home kernel: [76478.454250] [<c0145185>] warn_slowpath_null+0x15/0x20
Feb 2 12:51:38 kovalev-home kernel: [76478.454254] [<c04d2439>] tcp_recvmsg+0xa49/0xb20
Feb 2 12:51:38 kovalev-home kernel: [76478.454259] [<c0492433>] sock_common_recvmsg+0x43/0x60
Feb 2 12:51:38 kovalev-home kernel: [76478.454263] [<c0492...

[I Kovalev (iakovalev)]

System information in addition to my previous post attached:
(https://bugs.launchpad.net/ubuntu/+source/linux/+bug/147464/comments/165)

[I Kovalev (iakovalev)]

System information in addition to my previous post attached:
(https://bugs.launchpad.net/ubuntu/+source/linux/+bug/147464/comments/165)

[gr8viju (gr8viju)]

Hi I am also facing similar issues since, I am using Karmic 32 bit 2.6.31.19-Generic. The system freezes and nothing can be done except hard reboot. I have Intel 5150 WiMax card. This happens mostly using torrent clients and or with firefox with flash playing on it. Getting frustrated with it. If I use more than 10 files for downloading it happens within 5-10 mins.

Yesterday I switched to X86_64 bit version of Karmic 2.6.31.19-Generic. Will post the recent results.

[gr8viju (gr8viju)]

In X86_64 bit version also I am getting system hang. One torrent with speed more than 450KBps and one can see the result, This has become very annoying.

[I Kovalev (iakovalev)]

Have being using Transmission (1.75 to 1.91) for over a year in Ubuntu 9.04 to 9.10 on my Toshiba Satellite L40 laptop connected to the Internet via built-in Atheros AR5007EG wireless card installed in mini-PCI. While using Transmission I discovered that downloading from large swarms at high speed resulted in regular system crash (kernel panic) within 30 min to 1 hour. Recently it turned out the issue is system-wide, i.e. any heavy network activity triggers system crash. My search through Linux support community made me think that a bug in Linux kernel which causes i/o stalls in some cases (seems has been fixed in kernel 2.6.32) could be responsible for it.

Recently however I have found a hardware problem in my laptop. The PCI chip (not the CPU!) lacks proper cooling due to limitations in laptop design, mechanical defect in its radiator and partial blocking of ventilation slits by dust. After fixing mechanical problems my system has become rock stable. Still I was able to trigger system crash by heavy i/o activity - while downloading at ~2.5 MBits/s over wireless and transferring lots of data (~10MByte/s) via IEEE1394 PCMCIA card simultaneously and for a prolonged time providing that a temperature of cooling air is above 24 deg C.

Problem solved.
Conclusion: before reporting a software problem, it should be wise to check all your hardware carefully...

[RussianNeuroMancer (russianneuromancer)]

This issue is still present in Lucid RC?

[I Kovalev (iakovalev)]

Well, in my case there was (and still is) a hardware issue - due to flaws in design the PCI chip on a motherboard (not CPU neither graphic card!) has become overheated during prolonged heavy wireless transfer by additional heat dissipated in nearby on-board inverters. I have to fix the issue completely (I am very close to that) before testing my system again.

[gr8viju (gr8viju)]

Well in my case I don't think it is a hardware issue. I am also using windows7 alongside and I use utorrent for almost 3 days in a row and it works fine without any hang, but in Ubuntu 9.10 it was an issue with download speed more than 500 kbps. Right now in Lucid, I am still to encounter this problem; it works fine with speed up to 1.5 Mbps. Will update on this soon.

[I Kovalev (iakovalev)]

I had experienced this problem (system becomes hardlocked during heavy network transfer) on Toshiba Satellite L4 laptop in Jaunty and Karmic. Later it turned out there was a hardware fault in one of on-board inverters. Just have my motherboard replaced and the issue disappeared completely! No more system hardlocks, neither error messages in system logs for several days.

Conclusion: while reporting software failure one has to be cautious about the conditions of hardware. It even worth it to ask a specialist when possible. These faults are not easily identifiable, and manifestation can be contr-intuitive.

[Thomas (thoms)]

I hade some frezes as well with Ubuntu 9.10 64-Bit and 10.04

The problem was as followed:

As I started to download (in my case it was by JDownloader) after a while the whole system freezes.

I checked My RAMs and one seemed to be faulty but as I changed it to a new one the problem was still up.

We I used one single 1024MB and it worked perfect (each of the stand alone)

As soon I entered the second one the freezing problem is back.

I've seen at some forums threads, they had problems with the same wireless card as I use (Netgear WN311B).

For addition the same Netgear card was running fine on an other PC with Ubuntu 32-Bit

I was using the Broadcom driver by the proprietary drivers.

Now I changed the Wireless-card to a plug and Play (out of the box support) and all the freezing problems are gone.

I don't think the card is faulty, Probably something with the RAM addresses?

I have no idea, but the fault can be rebuilt.

Informations about my actual Hardware:

Tower Yeong-Yang Midi-Tower YY-5707 Black, 350Watt, ATX

Mainboard ASUS P5QL-EPU, FSB1600, Intel P43 Chipset, SATA, PCI-E, GLAN

Processor Intel CORE2Quad Q8400, QUAD-Core,1333MHz, 4MB Kentsfield, 2.66GHz, SpeedStep, I64bit, NX

Memory 4069MB DDR2 PC-6400, (2x2048MB), 240Pin, 800MHz.

Harddisc 500GB, SATA-II, 7200rpm, 16MB Cache und ein paar weiter NTFS Partitionen

Graphiccard nVidia GF GT210, 512MB DDR2, TV-Out, DVI, HDMI

Wirelesscard now: TP-Link TL-WN851N

I hope my informations helping you to find the source of the freezing problem

[Nathan Adams (nadams)]

My system is locks up when an application attempts to copy a large file to an nfs mount. The system becomes unresponsive to the point that I cannot even reboot via the command line over SSH.

For example, when soundKonverter copies an ogg file from /tmp to /home/music (nfs share).

I am unable to kill the copy command:

nate 2791 1 0 10:32 ? 00:00:00 cp /tmp/kde-nate/soundkonverterkMTAC0.ogg /home/music/Foo Fighters/[Foo Fighters] One By One - 07. Halo.ogg

And I see lots of messages like this in /var/log/syslog:

May 22 10:44:31 nereidum kernel: [ 1680.390484] INFO: task cp:2791 blocked for more than 120 seconds.
May 22 10:44:31 nereidum kernel: [ 1680.390488] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
May 22 10:44:31 nereidum kernel: [ 1680.390492] cp D 00000000ffffffff 0 2791 2503 0x00000000
May 22 10:44:31 nereidum kernel: [ 1680.390501] ffff88012a457c48 0000000000000082 0000000000015bc0 0000000000015bc0
May 22 10:44:31 nereidum kernel: [ 1680.390508] ffff8801291331a0 ffff88012a457fd8 0000000000015bc0 ffff880129132de0
May 22 10:44:31 nereidum kernel: [ 1680.390516] 0000000000015bc0 ffff88012a457fd8 0000000000015bc0 ffff8801291331a0
May 22 10:44:31 nereidum kernel: [ 1680.390523] Call Trace:
May 22 10:44:31 nereidum kernel: [ 1680.390545] [<ffffffffa0cff2b0>] ? nfs_wait_bit_uninterruptible+0x0/0x20 [nfs]
May 22 10:44:31 nereidum kernel: [ 1680.390552] [<ffffffff8153eb87>] io_schedule+0x47/0x70
May 22 10:44:31 nereidum kernel: [ 1680.390573] [<ffffffffa0cff2be>] nfs_wait_bit_uninterruptible+0xe/0x20 [nfs]
May 22 10:44:31 nereidum kernel: [ 1680.390579] [<ffffffff8153f3df>] __wait_on_bit+0x5f/0x90
May 22 10:44:31 nereidum kernel: [ 1680.390587] [<ffffffff812b6234>] ? __lookup_tag+0x64/0x120
May 22 10:44:31 nereidum kernel: [ 1680.390608] [<ffffffffa0cff2b0>] ? nfs_wait_bit_uninterruptible+0x0/0x20 [nfs]
May 22 10:44:31 nereidum kernel: [ 1680.390615] [<ffffffff8153f488>] out_of_line_wait_on_bit+0x78/0x90
May 22 10:44:31 nereidum kernel: [ 1680.390622] [<ffffffff81085360>] ? wake_bit_function+0x0/0x40
May 22 10:44:31 nereidum kernel: [ 1680.390643] [<ffffffffa0cff29f>] nfs_wait_on_request+0x2f/0x40 [nfs]
May 22 10:44:31 nereidum kernel: [ 1680.390665] [<ffffffffa0d036af>] nfs_wait_on_requests_locked+0x7f/0xd0 [nfs]
May 22 10:44:31 nereidum kernel: [ 1680.390688] [<ffffffffa0d04aee>] nfs_sync_mapping_wait+0x9e/0x1a0 [nfs]
May 22 10:44:31 nereidum kernel: [ 1680.390711] [<ffffffffa0d04ed9>] nfs_write_mapping+0x79/0xb0 [nfs]
May 22 10:44:31 nereidum kernel: [ 1680.390733] [<ffffffffa0d04f47>] nfs_wb_all+0x17/0x20 [nfs]
May 22 10:44:31 nereidum kernel: [ 1680.390751] [<ffffffffa0cf3eba>] nfs_do_fsync+0x2a/0x60 [nfs]
May 22 10:44:31 nereidum kernel: [ 1680.390770] [<ffffffffa0cf4105>] nfs_file_flush+0x75/0xa0 [nfs]
May 22 10:44:31 nereidum kernel: [ 1680.390777] [<ffffffff8114051c>] filp_close+0x3c/0x90
May 22 10:44:31 nereidum kernel: [ 1680.390783] [<ffffffff81140627>] sys_close+0xb7/0x120
May 22 10:44:31 nereidum kernel: [ 1680.390790] [<ffffffff810131b2>] system_call_fastpath+0x16/0x1b

$ uname -a
Linux nereidum 2.6.32-22-generic #33-Ubuntu SMP Wed Apr 28 13...

[Nathan Adams (nadams)]

I just upgraded to linux 2.6.33.3 and no joy - soundKonverter is stuck trying to copy a file to a nfs mount.

[Jeremy Foshee (jeremyfoshee)]

Folks,
    I have closed this bug as WontFix due to the age of the original report. This bug has progressed in a nebulous form as to be undefinable and unfixable.

Any of you who believe you are affected by this particular bug, please file a new bug so that it can be addressed on it's merit.

Thanks!

~JFo