[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Strange server crash on woody with gallery/php





Bob Proulx wrote:

Urs Schroffeneger wrote:
<SNIP>

That would cause a lot of disk I/O all of a sudden at that time.

<SNIP>

In addition to the disk I/O the rescaling would cause a lot of CPU
activity.

Indeed it is.

And there comes the problem: at some point on the process, the server turns off. It doesn't shut down, it turns off. Nothing in the logs:
syslog has some " -- MARK -- " and then "syslog restart"

Off as in the power supply is switched off?  Or off as in the kernel
halted but power remained on?

seems to be like the power supply is switched off

it doesn't seem to be the rescaling program, the bug comes some times before rescaling, some times after. Gallery has some debugging messages, so I tried some of the shell comands run to rescale, and had no problems doing so. Only thing I know is that the process is a little processor and disk intensive.

Although gallery may be triggering the problem I can't see how this
could be a gallery problem.  If you are causing a kernel panic or
otherwise crashing the machine then it is a kernel issue.
There isn't a kernel panic message on the console.

I have only access to the machine by ssh and web interface, so it's not really easy to test and restart the machine after a crash :-D

That makes things very tricky.
Yes, and travelling 300km is kinda long to press on a power switch

You say the machine is turning itself off.  That makes me think an APM
or ACPI function to turn the power supply off.  But perhaps it is just
crashing the kernel or halting the kernel and you are using a
vernacular description.  So there appear to be four possibilities and
all of them involve the kernel.

What version of the kernel have you installed?  You say Debian woody
which would mean to me the kernel-image-2.4.18-k7 or other
architecture specific package.  Then I would have loaded the specific
modules in /etc/modules that I would need for the machine such as
networking.  Check those for being reasonable.  Cross check that with
what is shown from /proc/cpuinfo for your cpu.  Is it a reasonable
combination?  You may be running the wrong kernel for your cpu.

If you have 'apm' or 'acpi' in your /etc/modules I would consider
removing those to remove the ability of the machine to turn itself
off.  But be careful since changes there might make the machine
unbootable and you would need console access to recover.

Bob
Here are some additional infos...

Artis:/etc# cat modules
# /etc/modules: kernel modules to load at boot time.
#
# This file should contain the names of kernel modules that are
# to be loaded at boot time, one per line.  Comments begin with
# a "#", and everything on the line after them are ignored.

usb-uhci
input
usbkbd
keybdev
busmouse
radeon
busmouse
radeon
ip_tables
Artis:/etc# uname -a
Linux Artis 2.4.18-bf2.4 #1 Son Apr 14 09:53:28 CEST 2002 i686 unknown
Artis:/etc# lsmod
Module                  Size  Used by    Not tainted
ip_tables              10432   0  (unused)
radeon                 92472   0  (unused)
busmouse                2912   0  (unused)
keybdev                 1664   0  (unused)
usbkbd                  2848   0  (unused)
input                   3072   0  [keybdev usbkbd]
usb-uhci               20708   0  (unused)
usbcore                48032   0  [usbkbd usb-uhci]
Artis:/etc# cat /proc/cpuinfo
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 15
model           : 2
model name      : Intel(R) Celeron(R) CPU 2.00GHz
stepping        : 7
cpu MHz         : 2004.448
cache size      : 20 KB
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 2
wp              : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm
bogomips        : 3997.69

Artis:/etc#

I think it's the package kernel-image-2.4.18-bf2.4 from the standard installation CD, so there isn't a lot of modules in it, everything is compiled in. I could try installing a new kernel, but I don't really like it, through ssh; could an 'apt-get install kernel-image-2.4.18-1-686' do it, (it does basicly contain all the drivers that where compiled in in the installation kernel, right ?) or do I risk some difficulties with missing modules? I only need a working network, the rest, I could fix from here. I will consider that one

So switching to another php version wouldn't resolve it. But i'm puzzled on the fact that to much disk activity could turn of the computer...

thanks

urs



Reply to: