2.6.13-rcX really this bad ?

2.6.13-rcX really this bad ?

[Danny ter Haar]

I've posted a couple of times than my newsserver is not stable
with any 2.6.13-rcX kernels.
Last kernel that survived is 2.6.12-mm1 (18+days)
Of course i can just stick with that kernel, but i thought it would
be wise to live on the edge and run a reasonable loaded server with
the latest/greatest. This ends in disaster though...

These are last uptimes, all reset by panic on reset.

reboot   system boot  2.6.13-rc6-git5  Sun Aug 14 02:56    (09:07)
reboot   system boot  2.6.13-rc6-git4  Fri Aug 12 16:50    (1+19:13)
reboot   system boot  2.6.13-rc6-git3  Fri Aug 12 12:32    (04:16)
reboot   system boot  2.6.13-rc6-git3  Thu Aug 11 13:18    (1+03:29)
reboot   system boot  2.6.13-rc6-git2  Thu Aug 11 12:40    (00:35)
reboot   system boot  2.6.13-rc6-git1  Wed Aug 10 13:24    (23:52)
reboot   system boot  2.6.13-rc6       Wed Aug 10 02:46    (10:35)
reboot   system boot  2.6.13-rc6       Mon Aug  8 16:52    (1+20:29)
reboot   system boot  2.6.13-rc6       Sun Aug  7 22:07    (2+15:14)

Since i got no feedback on my previous posts, i either bring it 
the wrong way, or people don't care and i ought to shut up.
I think however that just before releasing a new stable kernel these
kind of feedback could be healthy to ironout some bugs.

For now i'll switch back to 2.6.12-mm1 and follow this mailling list
and try again when i the problems seem to have been solved.

Danny

Re: 2.6.13-rcX really this bad ?

[Alexander Nyberg]

On Sun, Aug 14, 2005 at 10:10:18AM +0000 Danny ter Haar wrote:

> I've posted a couple of times than my newsserver is not stable
> with any 2.6.13-rcX kernels.
> Last kernel that survived is 2.6.12-mm1 (18+days)
> Of course i can just stick with that kernel, but i thought it would
> be wise to live on the edge and run a reasonable loaded server with
> the latest/greatest. This ends in disaster though...
>
> Since i got no feedback on my previous posts, i either bring it 
> the wrong way, or people don't care and i ought to shut up.
> I think however that just before releasing a new stable kernel these
> kind of feedback could be healthy to ironout some bugs.
> 

Is the machine running X? We need some output from it so we can debug
what's going on, the info should be printed to the console. It would
be great if you could run the latest kernel and see if you get any
output. Also add nmi_watchdog=2 to the boot command line.

You can also set up a serial console or netconsole to capture the output
from the server with the help of another machine, described in
Documentation/serial-console.txt
Documentation/networking/netconsole.txt

Re: 2.6.13-rcX really this bad ?

[Alexey Dobriyan]

On Sun, Aug 14, 2005 at 01:43:18PM +0200, Alexander Nyberg wrote:
> On Sun, Aug 14, 2005 at 10:10:18AM +0000 Danny ter Haar wrote:
> > I've posted a couple of times than my newsserver is not stable
> > with any 2.6.13-rcX kernels.
> > Last kernel that survived is 2.6.12-mm1 (18+days)

> Is the machine running X? We need some output from it so we can debug
> what's going on, the info should be printed to the console. It would
> be great if you could run the latest kernel and see if you get any
> output. Also add nmi_watchdog=2 to the boot command line.
> 
> You can also set up a serial console or netconsole to capture the output
> from the server with the help of another machine, described in
> Documentation/serial-console.txt
> Documentation/networking/netconsole.txt

Danny, could you please go to http://bugme.osdl.org/show_bug.cgi?id=4982
and fill new info there.

Alexander, could you please forward this email to Danny if you don't get
bounces from him. I got this when tried to reply that his bug was sucked
into bugzilla.
----------------------------------------------------------------------------
Delivery to the following recipient failed permanently:

     dth@picard.cistron.nl

Technical details of permanent failure:
TEMP_FAILURE: Could not initiate SMTP conversation with any hosts:
[bitbucket.cistron.nl (100): Connection timed out]

Re: 2.6.13-rcX really this bad ?

[Danny ter Haar]

Alexey Dobriyan  <adobriyan@gmail.com> wrote:
>> Is the machine running X? We need some output from it so we can debug
>> what's going on, the info should be printed to the console. It would
>> be great if you could run the latest kernel and see if you get any
>> output. Also add nmi_watchdog=2 to the boot command line.

No, it's not running X. It's running debian (amd64)
I allready posted more details http://newsgate.newsserver.nl/kernel/ 
including problems with SCSI.
_my_ guess it's an IRQ problem (acpi related) ?

>> You can also set up a serial console or netconsole to capture the output
>> from the server with the help of another machine, described in
>> Documentation/serial-console.txt
>> Documentation/networking/netconsole.txt

As stated before, it's connected to a sercon, but not with capture
capabilities. I've a program that captured a few panics/resets

>Danny, could you please go to http://bugme.osdl.org/show_bug.cgi?id=4982
>and fill new info there.

I certainly will!

>Alexander, could you please forward this email to Danny if you don't get
>bounces from him. I got this when tried to reply that his bug was sucked
>into bugzilla.

>     dth@picard.cistron.nl
>Technical details of permanent failure:
>TEMP_FAILURE: Could not initiate SMTP conversation with any hosts:
>[bitbucket.cistron.nl (100): Connection timed out]

Ouch, wil look into this ...

Danny
dth@cistron.nl