Hi all,

I have installed Slackware 10.0 on many computers and all works fine.
However, on one computer I have a strange issue. It's about Proftpd.
If it runs standalone, everything is fine. But if I run it throught
inetd, as soon as I try to transfer many files (= about 200 files
making about 100MB in size), the entire system hangs. I have screen,
but I can move mouse, keyboard doesn't respond, I can't ping the
computer anymore... nothing helps, so I have to press the reset button.

Slackware is a fresh 10.0 install out-of-the box, without any
modifications. The computer is brand new machine:

Intel P4 @3Ghz
512MB RAM
160GB HDD
....etc.

I really don't understand what is happening, nor where to look for
problems. I did search /var/log/messages, but found nothing.

Has anyone experienced something like this?

--
Milan Babuskov
http://njam.sourceforge.net

Milan Babuskov <milanb.ukloniovo@kms.co.yu> wrote:
> But if I run it throught inetd, as soon as I try to transfer many files
> (= about 200 files making about 100MB in size), the entire system hangs.
> I have screen, but I can move mouse, keyboard doesn't respond, I can't
> ping the computer anymore... nothing helps, so I have to press the reset
> button.

This kind of hang is most likely caused by a broken driver or some broken
hardware. Things to check:

1) Does memtest86 have anything to say about your RAM?

2) Would it help if you replace your NIC?

3) Double-check that your kernel and the module for your NIC is compiled
with the same configuration. You should also check this for your other
modules. If you are running an SMP kernel with non-SMP modules you could
get more or less random hangups.

regards Henrik

--
The address in the header is only to prevent spam. My real address is:
hc2(at)uthyres.com Examples of addresses which go to spammers:
info@k-software.biz svar@webtelevision.se info@webrider.ru root@localhost

Henrik Carlqvist wrote:
>>But if I run it throught inetd, as soon as I try to transfer many files
>>(= about 200 files making about 100MB in size), the entire system hangs.
>>I have screen, but I can move mouse, keyboard doesn't respond, I can't
>>ping the computer anymore... nothing helps, so I have to press the reset
>>button.
>
> This kind of hang is most likely caused by a broken driver or some broken
> hardware. Things to check:
>
> 1) Does memtest86 have anything to say about your RAM?

I'll check this out.

> 2) Would it help if you replace your NIC?

This seems like a potential problem. NIC is integrated on mainboard,
and only works when hotplug is enables. When the system is booting,
the Starting hotplug part takes a lot of time (~1min). Total boot time
is under 2 mins, so you get the picture.

> 3) Double-check that your kernel and the module for your NIC is compiled
> with the same configuration. You should also check this for your other
> modules. If you are running an SMP kernel with non-SMP modules you could
> get more or less random hangups.

I used the stock kernel and modules from CDs, didn't compile anything
yet (I machine is not a week old).

Anyway, I still don't understand why it works ok when proftpd is
started manually as standalone service. Maybe I was just lucky?

Anyway, thank you for the ideas, I'll look into it.

Regards,

--
Milan Babuskov
http://abrick.sourceforge.net

On Wed, 23 Feb 2005 20:12:41 +0100, Milan Babuskov <milanb.ukloniovo@kms.co.yu> wrote:

>Hi all,
>
>I have installed Slackware 10.0 on many computers and all works fine.
>However, on one computer I have a strange issue. It's about Proftpd.
>If it runs standalone, everything is fine. But if I run it throught
>inetd, as soon as I try to transfer many files (= about 200 files
>making about 100MB in size), the entire system hangs.

I'll assume you also told proftpd it was no longer running
standalone?
....
>I really don't understand what is happening, nor where to look for
>problems. I did search /var/log/messages, but found nothing.

And look at other logs, ('ls -lrt' to see most recent) and try
telling syslogd to not buffer log writes, more info may be
available next time.

But if you are sure the error ONLY happens after a config change,
the error would have to be in that config change, no?

Make sure correct NIC driver loaded -- the startup scripts do
make assumptions that may no longer apply.

Compile kernel for your hardware, confirm system operation in
chunks: memory, filesystem and networking.

Cheers,
Grant.

GrantC wrote:
>>I have installed Slackware 10.0 on many computers and all works fine.
>>However, on one computer I have a strange issue. It's about Proftpd.
>>If it runs standalone, everything is fine. But if I run it throught
>>inetd, as soon as I try to transfer many files (= about 200 files
>>making about 100MB in size), the entire system hangs.
>
> I'll assume you also told proftpd it was no longer running
> standalone?

Yes. Actually, I first tried to run it via inetd. And system hanged
few times. I started to thing what was I doing, and got down to
proftpd. After I removed it from inet, and run manually, everything
worked fine. I tried to get it back to inet, and it hangs again. Of
course I did make sure it is exclusive inet or standalone (it wouldn't
be able to open port anyway).

>>I really don't understand what is happening, nor where to look for
>>problems. I did search /var/log/messages, but found nothing.
>
> And look at other logs, ('ls -lrt' to see most recent)

Since I have to reboot before I review them, the times get reset. I
guess I'll have to boot from CD, mount the partition and then look at it.

> and try
> telling syslogd to not buffer log writes, more info may be
> available next time.

Hm, I didn't know that there is such option. Thanx.

> But if you are sure the error ONLY happens after a config change,
> the error would have to be in that config change, no?

Yes. But it is really strange. I connect to proftpd, it lets me list
directories, etc. When I copy a lot of files, it starts copying,
copies some files and then suddenly the entire system hangs. No ping,
mouse and keyboard don't respond...

> Make sure correct NIC driver loaded -- the startup scripts do
> make assumptions that may no longer apply.

The hotplug script loads it. NIC is integrated in mainboard. I may
have to check if there are different drivers for it. Is it possible to
have mulitple drivers for the same thing?

> Compile kernel for your hardware, confirm system operation in
> chunks: memory, filesystem and networking.

I was hoping not having to do that. It is not my home machine, but a
friend's at remote location.

Thanx for the ideas.

Regards,
--
Milan Babuskov
http://abrick.sourceforge.net

On Fri, 25 Feb 2005 18:48:38 +0100, Milan Babuskov <milanb.ukloniovo@kms.co.yu> wrote:
....
>> And look at other logs, ('ls -lrt' to see most recent)
>
>Since I have to reboot before I review them, the times get reset. I
>guess I'll have to boot from CD, mount the partition and then look at it.

If they not written, I don't think that will help,
lockup --> no log write flush to disk
....
>
>Yes. But it is really strange. I connect to proftpd, it lets me list
>directories, etc. When I copy a lot of files, it starts copying,
>copies some files and then suddenly the entire system hangs. No ping,
>mouse and keyboard don't respond...
Sounds like hardware IRQ lockup --> wrong driver.
>
>> Make sure correct NIC driver loaded -- the startup scripts do
>> make assumptions that may no longer apply.
>
>The hotplug script loads it. NIC is integrated in mainboard. I may
>have to check if there are different drivers for it. Is it possible to
>have mulitple drivers for the same thing?

Yep, depends on the NIC chipset, check /etc/hotplug/blacklist
to see if there is an alternate driver for your situation.

I've hit two: eepro100 vs e100, and tulip vs de4x5.

Cheers,
Grant.

GrantC <grant_nospam@dodo.com.au> wrote:
>>> And look at other logs, ('ls -lrt' to see most recent)

>>Since I have to reboot before I review them, the times get reset. I
>>guess I'll have to boot from CD, mount the partition and then look at it.

> If they not written, I don't think that will help,
> lockup --> no log write flush to disk

When trying to track hard locks in the kernel I usually configure syslogd
to log to /dev/tty12 and then make sure that tty is active and not blanked
when the machine hangs.

Try this:
echo hello world > /dev/tty12
(then press ctrl-alt-f12 to view the result)

If I remember right the following line should then go to /etc/syslog.conf:
*.* /dev/tty12

regards Henrik
--
The address in the header is only to prevent spam. My real address is:
hc2(at)uthyres.com Examples of addresses which go to spammers:
info@k-software.biz svar@webtelevision.se info@webrider.ru root@localhost

On Sat, 26 Feb 2005 19:03:15 +0100, Henrik Carlqvist <Henrik.Carlqvist@deadspam.com> wrote:

>GrantC <grant_nospam@dodo.com.au> wrote:
....
>> If they not written, I don't think that will help,
>> lockup --> no log write flush to disk
>
>When trying to track hard locks in the kernel I usually configure syslogd
>to log to /dev/tty12 and then make sure that tty is active and not blanked
>when the machine hangs.
>
>Try this:
>echo hello world > /dev/tty12
>(then press ctrl-alt-f12 to view the result)
>
>If I remember right the following line should then go to /etc/syslog.conf:
>*.* /dev/tty12
Hi Henrik,
Didn't think to suggest testing at CLI to eliminate GUI interference,
and do we have a dozen terminals by default? That's interesting, Alt-12
shows hello world, no login prompt... Learn something every day

Cheers,
Grant.

GrantC <grant_nospam@dodo.com.au> wrote:
>>Try this:
>>echo hello world > /dev/tty12
>>(then press ctrl-alt-f12 to view the result)

> and do we have a dozen terminals by default? That's interesting, Alt-12
> shows hello world, no login prompt... Learn something every day

In Slackware /etc/inittab is configured to have login prompts on the first
6 virtual consoles only. When X is started it uses the first free virtual
console which usually is tty7. It is possible to start more than one X
session with something like this:

startx -- :1

The above command will start a new X session on the next free virtual
console which will be tty8 if we already have an X with DISPLAY :0 on
tty7. The argument ":1" tells X to use DISPLAY :1 so you will be able to
put windows on that X with "export DISPLAY=:1". As you now understand it
is possible to have more than two X running on the same machine
simultaneously. I haven't tried what happens if more than 6 X would be
started, I don't know if there is any tty13, and if so, how you would get
to that tty.

regards Henrik
--
The address in the header is only to prevent spam. My real address is:
hc2(at)uthyres.com Examples of addresses which go to spammers:
info@k-software.biz svar@webtelevision.se info@webrider.ru root@localhost

Milan Babuskov wrote:
> I have installed Slackware 10.0 on many computers and all works fine.
> However, on one computer I have a strange issue. It's about Proftpd. If
> it runs standalone, everything is fine.

Later I have noticed that it hangs even when running standalone.

> as soon as I try to transfer many files (= about 200 files making about
> 100MB in size), the entire system hangs. I have screen, but I can move
> mouse, keyboard doesn't respond, I can't ping the computer anymore...
> nothing helps, so I have to press the reset button.

The problem was in r8169 NIC module in kernel 2.4.26. I upgraded to 2.4.29
today, and everything works fine.

Thanx to everyone for tips how to solve this.

--
Milan Babuskov
http://njam.sourceforge.net