Re: [gentoo-amd64] System crashes when idle [SOLUTION(?)]

2007-06-08 Thread Michael Ulm

Michael Ulm wrote:

Hi,

I seem to get system crashes on my machine when it is idle for
some time. Here is a typical crashing session:

I leave the computer on idle for an hour or so. When I return,
one or more applications have crashed. dmesg will show a segfault
on some apparently random location, and the system will remain
highly unstable. Usually, even halting the system will not be
possible. The next reboot will produce several errors on the hard
disk, which usually need manual intervention. After some repairing,
the system then runs stable again.

Can this be a software problem? I tried to disable all sleep/hibernate
functionality in KDE, but maybe I'm missing something there.

Maybe it's the hardware, but what kind of hardware issue could cause
these symptoms?


I installed Ubuntu on a second disk and it didn't crash. After a lengthy
process of comparing setups, I think I found the problem.

In xorg.conf, in the monitor section, I was missing the line
Options=DPMS

I'll still have to research how leaving out this option could lead to
crashes, so I appreciate any insight this group may give.

Regards,

Michael

--
Michael Ulm
RD Team
ISIS Information Systems Austria
tel: +43 2236 27551-542, fax: +43 2236 21081
e-mail: [EMAIL PROTECTED]
Visit our Website: www.isis-papyrus.com

---
This e-mail is only intended for the recipient and not legally
binding. Unauthorised use, publication, reproduction or
disclosure of the content of this e-mail is not permitted.
This email has been checked for known viruses, but ISIS accepts
no responsibility for malicious or inappropriate content.
--- 
--

[EMAIL PROTECTED] mailing list



Re: [gentoo-amd64] System crashes when idle [SOLUTION(?)]

2007-06-08 Thread Sebastian Redl
Michael Ulm wrote:
 I installed Ubuntu on a second disk and it didn't crash. After a lengthy
 process of comparing setups, I think I found the problem.

 In xorg.conf, in the monitor section, I was missing the line
 Options=DPMS

 I'll still have to research how leaving out this option could lead to
 crashes, so I appreciate any insight this group may give.
http://en.wikipedia.org/wiki/VESA_Display_Power_Management_Signaling


Since practically every monitor supports this, and since it's
practically always enabled, it may be that other code paths of X.org are
insufficiently tested, causing the server to crash when it wants to put
the monitor into a power saver mode. That's my best guess.

Sebastian Redl
-- 
[EMAIL PROTECTED] mailing list



Re: [gentoo-amd64] System crashes when idle [SOLUTION(?)]

2007-06-08 Thread Peter Humphrey
On Friday 08 June 2007 10:19:08 Sebastian Redl wrote:
 Michael Ulm wrote:
  In xorg.conf, in the monitor section, I was missing the line
  Options=DPMS
 
  I'll still have to research how leaving out this option could lead to
  crashes, so I appreciate any insight this group may give.

 Since practically every monitor supports this, and since it's
 practically always enabled, it may be that other code paths of X.org are
 insufficiently tested, causing the server to crash when it wants to put
 the monitor into a power saver mode. That's my best guess.

My experience doesn't match this. I've had to remove the DPMS option from my 
xorg.conf to prevent lockups. Those only happened if I left the KDM login 
screen up until the screen blanker cut in, but then the only way out was a 
hard reset.

Apart from wanting my machine to do only what I tell it to, and I'm quite 
capable of switching my monitor off when I'm not using it, I think it's 
dangerous for any system to make assumptions about what facilities will be 
required in any particular case.

My hardware is an nVidia GeForce 7300 GS driving an Iiyama AU4831D screen, 
and the X nv driver comes from x11-base/xorg-x11-7.2. Or sometimes I play 
with the nVidia closed-source drivers, but as far as I know that doesn't 
affect the lockups.

-- 
Rgds
Peter Humphrey
Linux Counter 5290, Aug 93
-- 
[EMAIL PROTECTED] mailing list



Re: [gentoo-amd64] System crashes when idle [SOLUTION(?)]

2007-06-08 Thread Dustin C. Hatch
I have been experiencing this problem for quite some time now, but I
have found that when I use gentoo-sources-2.6.20-r4 instead of other
kernels, the problem seems to go away.

Last night, I was trying 2.6.21-r2 to see if I still experienced the
problem.  I had started a lengthy copy process before going to bed. 
When I woke this morning, the computer was in fact locked up.  Like
always, the cursor still moved around the screen when I moved the mouse,
but button clicks and keyboard keystrokes had no effect.  Again, I had
to use the Magic SysRq sequence to reboot the machine.  When it had
rebooted, back to 2.6.20-r4, I restarted rsync to finish the copy
process.  To my surprise, the copy was already finished.

I should point out that every time I experience this, I try
unsuccessfully to ssh into my computer.  The computer does not respond
to any network traffic I initiate, SSH, ICMP, HTTP, etc.

I have an nVidia GeForce 6600 running the proprietary nvidia driver. 
I h ave no DPMS option in my xorg.conf, and xscreensaver is running,
though set to never display a screen saver.

Dustin C. Hatch
http://www.dchweb.com



Peter Humphrey wrote:
 On Friday 08 June 2007 10:19:08 Sebastian Redl wrote:
   
 Michael Ulm wrote:
 
 In xorg.conf, in the monitor section, I was missing the line
 Options=DPMS

 I'll still have to research how leaving out this option could lead to
 crashes, so I appreciate any insight this group may give.
   
 Since practically every monitor supports this, and since it's
 practically always enabled, it may be that other code paths of X.org are
 insufficiently tested, causing the server to crash when it wants to put
 the monitor into a power saver mode. That's my best guess.
 

 My experience doesn't match this. I've had to remove the DPMS option from my 
 xorg.conf to prevent lockups. Those only happened if I left the KDM login 
 screen up until the screen blanker cut in, but then the only way out was a 
 hard reset.

 Apart from wanting my machine to do only what I tell it to, and I'm quite 
 capable of switching my monitor off when I'm not using it, I think it's 
 dangerous for any system to make assumptions about what facilities will be 
 required in any particular case.

 My hardware is an nVidia GeForce 7300 GS driving an Iiyama AU4831D screen, 
 and the X nv driver comes from x11-base/xorg-x11-7.2. Or sometimes I play 
 with the nVidia closed-source drivers, but as far as I know that doesn't 
 affect the lockups.

   


[gentoo-amd64] System crashes when idle

2007-05-29 Thread Michael Ulm

Hi,

I seem to get system crashes on my machine when it is idle for
some time. Here is a typical crashing session:

I leave the computer on idle for an hour or so. When I return,
one or more applications have crashed. dmesg will show a segfault
on some apparently random location, and the system will remain
highly unstable. Usually, even halting the system will not be
possible. The next reboot will produce several errors on the hard
disk, which usually need manual intervention. After some repairing,
the system then runs stable again.

Can this be a software problem? I tried to disable all sleep/hibernate
functionality in KDE, but maybe I'm missing something there.

Maybe it's the hardware, but what kind of hardware issue could cause
these symptoms?

Any ideas anyone?

Thanks,

Michael

--
Michael Ulm
RD Team
ISIS Information Systems Austria
tel: +43 2236 27551-542, fax: +43 2236 21081
e-mail: [EMAIL PROTECTED]
Visit our Website: www.isis-papyrus.com

---
This e-mail is only intended for the recipient and not legally
binding. Unauthorised use, publication, reproduction or
disclosure of the content of this e-mail is not permitted.
This email has been checked for known viruses, but ISIS accepts
no responsibility for malicious or inappropriate content.
--- 
--

[EMAIL PROTECTED] mailing list



Re: [gentoo-amd64] System crashes when idle

2007-05-29 Thread Peter Humphrey
On Tuesday 29 May 2007 07:59:42 Michael Ulm wrote:
 Hi,

 I seem to get system crashes on my machine when it is idle for
 some time. Here is a typical crashing session:
[...]
 Maybe it's the hardware, but what kind of hardware issue could cause
 these symptoms?

 Any ideas anyone?

A couple. Either RAM or hard disk would be my first suspect. You can check 
the RAM best by compiling any large package, such as gcc itself. Memtest 
and its like can only find the most obvious faults and in my opinion aren't 
worth their salt - if they can find a fault, it will already be obvious to 
you.

Secondly, with the advent of SATA I'm suspecting a new class of fault in 
hard disks. Not only do we have the traditional transfer errors, but now I 
suspect it's possible for some faults to allow voltage spikes to be imposed 
on the line driver and cause random errors in other subsystems. I may be 
wrong, but I've had one experience of this already, and I'm currently 
working to eliminate another disk as the cause of weird problems I'm 
having.

Oh, and there's always the power supply. If that goes wonky it can cause all 
manner of problems. You could well suspect it if problems don't occur until 
after the system's been running for quite a while.

-- 
Rgds
Peter Humphrey
Linux Counter 5290, Aug 93
-- 
[EMAIL PROTECTED] mailing list



Re: [gentoo-amd64] System crashes when idle

2007-05-29 Thread Michael Ulm

Peter Humphrey wrote:

On Tuesday 29 May 2007 07:59:42 Michael Ulm wrote:

Hi,

I seem to get system crashes on my machine when it is idle for
some time. Here is a typical crashing session:

[...]

Maybe it's the hardware, but what kind of hardware issue could cause
these symptoms?

Any ideas anyone?


A couple. Either RAM or hard disk would be my first suspect. You can check 
the RAM best by compiling any large package, such as gcc itself. Memtest 
and its like can only find the most obvious faults and in my opinion aren't 
worth their salt - if they can find a fault, it will already be obvious to 
you.


Secondly, with the advent of SATA I'm suspecting a new class of fault in 
hard disks. Not only do we have the traditional transfer errors, but now I 
suspect it's possible for some faults to allow voltage spikes to be imposed 
on the line driver and cause random errors in other subsystems. I may be 
wrong, but I've had one experience of this already, and I'm currently 
working to eliminate another disk as the cause of weird problems I'm 
having.


Oh, and there's always the power supply. If that goes wonky it can cause all 
manner of problems. You could well suspect it if problems don't occur until 
after the system's been running for quite a while.




Thank you for your suggestions. I have had no problems so far compiling my
Gentoo packages, and I often stressed my system with lengthy computations
without any issues. So I tend to think the RAM is good.

I thought that power supply issues would tend to occur under stress, not under
rest. But you are right of course, the problems may occur after some delay.
Unfortunately I've no idea how to test this.

I'll check the disk to see if something comes up there. I also plan to install
Ubuntu on some separate disk and see if the problem occurs there as well.

If these are indeed hardware related problems, I feel like abusing this group
(which is software related after all) with my problems. So, I'd also appreciate
suggestions which online community would be better suited for help with
debugging hardware issues.

Thanks,

Michael



--
Michael Ulm
RD Team
ISIS Information Systems Austria
tel: +43 2236 27551-542, fax: +43 2236 21081
e-mail: [EMAIL PROTECTED]
Visit our Website: www.isis-papyrus.com

---
This e-mail is only intended for the recipient and not legally
binding. Unauthorised use, publication, reproduction or
disclosure of the content of this e-mail is not permitted.
This email has been checked for known viruses, but ISIS accepts
no responsibility for malicious or inappropriate content.
--- 
--

[EMAIL PROTECTED] mailing list



Re: [gentoo-amd64] System crashes when idle

2007-05-29 Thread Dustin J. Mitchell
On Tue, May 29, 2007 at 08:42:10AM +0100, Peter Humphrey wrote:
 A couple. Either RAM or hard disk would be my first suspect. You can check 
 the RAM best by compiling any large package, such as gcc itself. Memtest 
 and its like can only find the most obvious faults and in my opinion aren't 
 worth their salt - if they can find a fault, it will already be obvious to 
 you.

FWIW, their value is in finding that the fault is with RAM, not with
anything else.  Compiling isn't going to narrow down the problem beyond
yep, the computer's broken :)

That said, that this happens when the system is quiet suggests it's not
the HD.

Dustin
-- 
[EMAIL PROTECTED] mailing list



Re: [gentoo-amd64] System crashes when idle

2007-05-29 Thread Joshua Hoblitt
On Tue, May 29, 2007 at 08:42:10AM +0100, Peter Humphrey wrote:
 A couple. Either RAM or hard disk would be my first suspect. You can check 
 the RAM best by compiling any large package, such as gcc itself. Memtest 
 and its like can only find the most obvious faults and in my opinion aren't 
 worth their salt - if they can find a fault, it will already be obvious to 
 you.

I've also found GCC to be a fairly effective system torture test.  Late
last year I had an issue where compiling GCC was the only relibable way
to reproduce an issue what we were seeing.  The symptoms were random
system hangs and believe it or not occasional segfaults.  It turned out
be bad L2 cache on the CPU.  Repeatedly emerging gcc eventually
generated a MCE log message allowing us to diagnose the problem.

-J

--


pgpBo19FaqybG.pgp
Description: PGP signature