Kyrian | 29 Jun 2006 00:24
Favicon

Re: TUX kernel panic - directory for log file doesn't exist...? (additional)

All,

Well, although it's not an SMP kernel *package* there were still reports 
of SMP tomfoolery in the output of dmesg, as follows:

...
found SMP MP-table at 000f4ca0
Using x86 segment limits to approximate NX protection
...
Enabling APIC mode:  Flat.  Using 1 I/O APICs
Using ACPI (MADT) for SMP configuration information
Allocating PCI resources starting at 40000000 (gap: 3ff00000:bed00000)
...

Therefore I have taken the (usually sane) step of disabling APIC and 
ACPI in the kernel parameters, and will see if this stops the random 
lockups.

This has had the effect of removing the latter section's reference to 
SMP, what that means in english, I have very little idea ;-)

I'm not going to risk restarting TUX again without a reboot because I 
don't want to lose the damn server.

This vexes me though. I've never had any problems with it before, and 
it's worked like a charm (on single processor boxes, or quad processor 
ones, although having just checked, both sites are running 2.4 kernels). 
Are there any other caveats I need to know about, or is it just with 2.6 
kernels?

(Continue reading)

Thierry de Montaudry | 8 Jul 2006 20:29
Picon

Tux kernel panic

Hi all,

Just started a new machine with CentOS 4.3 right off the box, minimum install to run Tux only, and I get some
kernal panic 
after a few hours (and a few 100000 hits).
See attached logs from /var/log/messages, crash1.log includes full dump... crash2.txt includes full
startup log.
The machine is a Intel P4-3.2Ghz with 1GB mem.
The only parameter that was changed for tux is:
net.tux.ignore_query = 2

IPV6 is disabled in the kernel... can it be a problem?

If there is a know problem with CentOS 4.3, will I get a chance to have a more stable tux with Fedora Core 4 or 5?

Regards and thanks by advance,

	Thierry

Attachment (crash1.log): application/octet-stream, 2408 bytes
Attachment (crash2.log): application/octet-stream, 23 KiB
_______________________________________________
tux-list mailing list
tux-list <at> redhat.com
https://www.redhat.com/mailman/listinfo/tux-list
Chris Davies | 8 Jul 2006 20:34
Favicon

Re: Tux kernel panic

I would guess you are using a kernel with SMT -- try using one without 
SMT/SMP and I bet the problem goes away.

Thierry de Montaudry wrote:
> Hi all,
> 
> Just started a new machine with CentOS 4.3 right off the box, minimum install to run Tux only, and I get some
kernal panic 
> after a few hours (and a few 100000 hits).
> See attached logs from /var/log/messages, crash1.log includes full dump... crash2.txt includes full
startup log.
> The machine is a Intel P4-3.2Ghz with 1GB mem.
Thierry de Montaudry | 10 Jul 2006 18:48
Picon

Re: Tux kernel panic

Hi,

Restarted the machine without SMP kernel... and got the exact same problem after a few hours.
Any clue?

Regards,

Thierry

On Sat, 08 Jul 2006 14:34:52 -0400, Chris Davies wrote:

>I would guess you are using a kernel with SMT -- try using one without
>SMT/SMP and I bet the problem goes away.
>
>Thierry de Montaudry wrote:
>> Hi all,
>>
>> Just started a new machine with CentOS 4.3 right off the box, minimum install to run Tux only, and I get some kernal panic
>> after a few hours (and a few 100000 hits).
>> See attached logs from /var/log/messages, crash1.log includes full dump... crash2.txt includes full startup log.
>> The machine is a Intel P4-3.2Ghz with 1GB mem.
>
>_______________________________________________
>tux-list mailing list
>tux-list <at> redhat.com
>https://www.redhat.com/mailman/listinfo/tux-list
>

_______________________________________________
tux-list mailing list
tux-list <at> redhat.com
https://www.redhat.com/mailman/listinfo/tux-list
Thierry de Montaudry | 8 Jul 2006 21:15
Picon

Re: Tux kernel panic

Hi Chris,

Thanks for the info. Just a pity to not be able to use the SMP side of the kernel and tux.
After I sent my mail, I discovered that the /etc/rc.d/ini.d/tux script has a bug and will report 0 CPUs instead of 2 could that create problems with tux?
If I execute the following line manually, I always get 0 as result.
NRCPUS=$(cat /proc/cpuinfo | grep '^processor[ 0-9]*:' | wc -l)

Regards,

Thierry

On Sat, 08 Jul 2006 14:34:52 -0400, Chris Davies wrote:

>I would guess you are using a kernel with SMT -- try using one without
>SMT/SMP and I bet the problem goes away.
>
>Thierry de Montaudry wrote:
>> Hi all,
>>
>> Just started a new machine with CentOS 4.3 right off the box, minimum install to run Tux only, and I get some kernal panic
>> after a few hours (and a few 100000 hits).
>> See attached logs from /var/log/messages, crash1.log includes full dump... crash2.txt includes full startup log.
>> The machine is a Intel P4-3.2Ghz with 1GB mem.
>
>_______________________________________________
>tux-list mailing list
>tux-list <at> redhat.com
>https://www.redhat.com/mailman/listinfo/tux-list
>

_______________________________________________
tux-list mailing list
tux-list <at> redhat.com
https://www.redhat.com/mailman/listinfo/tux-list

Gmane