[vox-tech] Redhat 8.0 shutdown does not power off.

Jim Angstadt vox-tech@lists.lugod.org
Thu, 5 Jun 2003 10:11:23 -0700 (PDT)


--- Mike Simons <msimons@moria.simons-clan.com>
wrote:
> On Wed, Jun 04, 2003 at 04:57:42PM -0700, Jim
> Angstadt wrote:
> > --- Mike Simons
> <msimons@moria.simons-clan.com> wrote:
> > > - using magic sysrq you can trigger the
> power off kernel function
> > >   to see if the kernel method works
> correctly on a given box.  With
> > >   'q', as in s-u-q, instead of s-u-b.
> > 
> > I read, but didn't understand, the earlier
> > threads on using magic sysrq keys.  I'll go
> back
> > over those threads.
> 
> Jim,
> 
>   Skim over
> /usr/src/linux/Documentation/sysrq.txt
> 
>   Basically switch to a command prompt so you
> can see what is happening,
> then hold down alt-SysRQ and press 's' you
> should see a message appear
> saying your file systems are being sync'ed.  if
> that happens hold down
> alt-SysRq and press 'u', you will see a message
> saying your filesystems
> are being remounted read-only (they will not
> need to be fsck'ed on next
> boot).  Then alt-SysRq and tap 'q', the kernel
> will call the kernel 
> function which is responsible for powering off
> the machine, you should
> see the power go off.
> 
> - Doing the above is meant to verify the kernel
> is correctly configured
>   for your BIOS... it should power off.  Try it
> a few times and report
>   back what happens.

I enabled sysrq by editing /etc/sysctl.conf and
then rebooted.  My normal state is to boot to the
command line (runlevel 3 in RH).

Without a login, I did Alt-SysRq-s and Emergency
Sync executed.  Then Alt-SysRq-u executed
Emergency Remount R/O.  Alt-SysRq-q gave me the
help menu.

I retried several times.  The last message was:

INIT: Id "1" respawning too fast: disabled for 5
minutes.

After about 5 minutes (I didn't time it) the same
message appeared.

I noted in sysreq.txt that there was no 'q'
listed as a command.

I'll wait until the "disabled..." messages stop. 
Then I'll shutdown, power off and power up to get
a clean starting point for more instructions.

Jim

> 
> > > - Paste output from:  grep MAGIC
> /boot/config*
> > 
> > ----- grep MAGIC /boot/config*
> > CONFIG_MAGIC_SYSRQ=y
> 
> Good, it is enabled.
> 
> > > > > - Paste output from:  lsmod
> > > > 
> > > > ----- lsmod
> > > > Module                  Size  Used by   
> > > Tainted: P  
> > > > nvidia               1593856  10 
> (autoclean)
> > > 
> > >   Hrmmm... tainted kernel.
> > 
> > The NVidia drivers are the only ones I
> > specifically downloaded and installed.  That
> was
> > done immediately upon original install.
> [...]
> > Maybe it's too soon to ask, but, are you
> thinking
> > that the original kernel was OK, and then
> later
> > was changed, not re-compiled, which resulted
> in
> > the current situation?
> 
>   It is unlikely that something magically
> changed from before to now.
> When you install kernel modules you don't
> actually change the kernel
> binaries, so you can "easily" go back to
> exactly what was before if you
> know what steps to take.
>   It is unlikely that the nvidia drivers are
> responsible... but it is 
> worth checking.
> 
> > If so, then an option would be to reinstall
> from CD, 
> [...]
> 
>   No, don't do a fresh install.  At least not
> now, it is likely to 
> change too many things.
> 
> 
>   There is a way to get Redhat to boot into
> text console mode (without
> starting X).  Based on what I see from lsmod,
> "autoclean", I think that
> the nvidia driver is being loaded by X when it
> starts... so I think
> simply not starting X will result in no
> loading.
>   There are about 20 different ways to stop X
> from starting... I'll
> mention two ways, report back if they both
> don't work right.  If 
> things have not changed in Redhat, then in
> /etc/inittab there 
> should be a line that says, something like:
>   id:3:initdefault:
> if you change the '3' to a '2', X should not
> start on the next reboot.
> 
> Otherwise if you move the X config file some
> place else, X will not
> start... the following two commands should
> handle that:
> ===
> mkdir /etx/X11/test
> mv /etc/X11/XF86Config* /etc/X11/test
> ===
> 
> to move them back into place:
> ===
> mv /etc/X11/test/* /etc/X11
> ===
> 
> 
>   The way to see 'not starting X' worked is to
> look at lsmod output
> again, there should be no 'nvidia' line... and
> the 'Tainted:' thing
> should say 'not tainted'.  If nvidia is missing
> but it says something
> else please paste the output of lsmod again and
> I'll try to figure out
> what else could be tainting.
> 
>   Once the kernel is non-tainted, try the
> normal "shutdown -h now"
> a few times and report back if there is any
> change from the unreliable
> behavior.
> 
>     Good Luck,
>       Mike
> 
> -- 
> GPG key:
> http://simons-clan.com/~msimons/gpg/msimons.asc
> Fingerprint: 524D A726 77CB 62C9 4D56  8109
> E10C 249F B7FA ACBE
> 

> ATTACHMENT part 2 application/pgp-signature 



__________________________________
Do you Yahoo!?
Yahoo! Calendar - Free online calendar with sync to Outlook(TM).
http://calendar.yahoo.com