Very Computer

One of the problems is that a warm boot never works.  It gets through
booting and throws up the graphic screen, and then the Xwindow login
window -- where I have no kbd or mouse operation -- not even ctl-alt-BKSP
or NumLock.  Okay --- but that's not why I'm posting.  Read on:

Being inside my 30-day window of 'free' "Premier Tech Support"
at RH, I've been going back and forth with them about things to
try --  m a n y  things.
The latest 'thing' was to update to kernel-2.4.9-21.

Did I mention I was a linux rookie for the most part?
My RH 6.1 was pretty solid (for me) and I never really
had to 'get my hands dirty'.

Well, now I fear my system is TOTALLY hosed.
I already had the new kernel -- downloaded by up2date
last week, but not installed.
I attempt to rpm -Uvh the kernel and I get "failed dependencies
 modutils >= 2.4.10 is needed".
OK.  I already had modutils-2.4.13-0.7.1.i386.rpm  -- again,
downloaded by up2date, but not installed.

I (aleady had) made a copy of my current vmlinuz-2.4.7-10 in /boot.
I installed the 2.4.13 modutils with rpm -- no problem(s).
I installed the 2.4.9 kernel with rpm -- no problem(s).
I updated /etc/lilo.conf.
I looked in /boot and initrd and vmlinuz were -2.4.9-21.
I re-booted.  
BIG PROBLEMS!
I get 1:
       "modprobe: Can't open dependencies file
               /lib/modules/2.4.7-10/modules.dep not found."
and at least one:
       "Finding module dependencies: depmod: can't open
               /lib/modules/2.4.7-10/modules.dep for writing"
and 3 or 4:
       "modprobe: Can't open dependencies file
               /lib/modules/2.4.7-10/modules.dep
                    (No such file or directory.)"

Not much of my system gets loaded.  My kbd and mouse are inop,
and I cannot telnet in from my OS/2 workstation on the lan.

I can boot into linux single from my rescue disk.  
grep tells me there's a "2.4.7-10" text string in /boot/map.
/boot/map is timestamped Feb 8  --  the date of my RH 6.1
to 7.2 upgrade.  I have no idea what /boot/map is, or is for,
or even _if_ it is part of my problem here.
"-2.4.7-10" is, of course, the version of my old vmlinuz and initrd.

I had made a copy of my old kernel according to the kernel
HOWTO directions I found on redhat.com  But, this being my
first kernel upgrade ever, I didn't know to make a backup of
my initrd to go with with my backup of the kernel.

I have *no* clue what I can do with the mess I now have to
get my system back.  I do not want to futz with it further
without 'guidance'. And there be *no* linux folks -- to my
knowledge -- within at least a 75 mile radius of me.

Can I fix/repair this?  
Or, do I just scrape it down to the bare metal?

Thanks for reading this way-too-long post.
Jonesy
--
  | Marvin L Jones       | jonz           |  W3DHJ   |  OS/2

  |   7,703' -- 2,345m   |   frontier.net |  DM68mn             SK

Quote:> Almost 2 weeks ago I did a RH 7.2 upgrade over my stable,
> fully functional RH 6.1.  I've been bleeding through one
> or both ears ever since.

> One of the problems is that a warm boot never works.  It gets through
> booting and throws up the graphic screen, and then the Xwindow login
> window -- where I have no kbd or mouse operation -- not even ctl-alt-BKSP
> or NumLock.  Okay --- but that's not why I'm posting.  Read on:

> Being inside my 30-day window of 'free' "Premier Tech Support"
> at RH, I've been going back and forth with them about things to
> try --  m a n y  things.
> The latest 'thing' was to update to kernel-2.4.9-21.

> Did I mention I was a linux rookie for the most part?
> My RH 6.1 was pretty solid (for me) and I never really
> had to 'get my hands dirty'.

> Well, now I fear my system is TOTALLY hosed.
> I already had the new kernel -- downloaded by up2date
> last week, but not installed.
> I attempt to rpm -Uvh the kernel and I get "failed dependencies
>  modutils >= 2.4.10 is needed".
> OK.  I already had modutils-2.4.13-0.7.1.i386.rpm  -- again,
> downloaded by up2date, but not installed.

> I (aleady had) made a copy of my current vmlinuz-2.4.7-10 in /boot.
> I installed the 2.4.13 modutils with rpm -- no problem(s).
> I installed the 2.4.9 kernel with rpm -- no problem(s).
> I updated /etc/lilo.conf.
> I looked in /boot and initrd and vmlinuz were -2.4.9-21.
> I re-booted.
> BIG PROBLEMS!
> I get 1:
>        "modprobe: Can't open dependencies file
>                /lib/modules/2.4.7-10/modules.dep not found."
> and at least one:
>        "Finding module dependencies: depmod: can't open
>                /lib/modules/2.4.7-10/modules.dep for writing"
> and 3 or 4:
>        "modprobe: Can't open dependencies file
>                /lib/modules/2.4.7-10/modules.dep
>                     (No such file or directory.)"

> Not much of my system gets loaded.  My kbd and mouse are inop,
> and I cannot telnet in from my OS/2 workstation on the lan.

> I can boot into linux single from my rescue disk.
> grep tells me there's a "2.4.7-10" text string in /boot/map.
> /boot/map is timestamped Feb 8  --  the date of my RH 6.1
> to 7.2 upgrade.  I have no idea what /boot/map is, or is for,
> or even _if_ it is part of my problem here.
> "-2.4.7-10" is, of course, the version of my old vmlinuz and initrd.

> I had made a copy of my old kernel according to the kernel
> HOWTO directions I found on redhat.com  But, this being my
> first kernel upgrade ever, I didn't know to make a backup of
> my initrd to go with with my backup of the kernel.

> I have *no* clue what I can do with the mess I now have to
> get my system back.  I do not want to futz with it further
> without 'guidance'. And there be *no* linux folks -- to my
> knowledge -- within at least a 75 mile radius of me.

> Can I fix/repair this?
> Or, do I just scrape it down to the bare metal?

> Thanks for reading this way-too-long post.
> Jonesy

    Boot the linux rescue CD.
    Copy the kernel RPM over to "/tmp" on the mounted root partition disk
(usually /mnt/sysimage).
    chroot to the mounted disk (chroot /mnt/sysimage)
    Install the RPM with "rpm -i /tmp/kernel*rpm"
    Edit lilo.conf or grub.conf appropriately.
    If you have a SCSI boot drive, you *MUST* do "mkinitrd" to generate a
new set of pre-loadable SCSI modules that you have listed in lilo.conf or
grub.conf
    Run lilo or grub with the new kernel. You may have to edit some
confusing and erroneous duplicate entries out of /etc/mtab in the mounted
partition to make this work correctly.
    Exit the chroot.
    Reboot the machine.

Quote:> I (aleady had) made a copy of my current vmlinuz-2.4.7-10 in /boot.
> I installed the 2.4.13 modutils with rpm -- no problem(s).
> I installed the 2.4.9 kernel with rpm -- no problem(s).
> I updated /etc/lilo.conf.

Quote:> I looked in /boot and initrd and vmlinuz were -2.4.9-21.
> I re-booted.  
> BIG PROBLEMS!

Quote:>> I installed the 2.4.13 modutils with rpm -- no problem(s).
>> I installed the 2.4.9 kernel with rpm -- no problem(s).
>> I updated /etc/lilo.conf.
>> I looked in /boot and initrd and vmlinuz were -2.4.9-21.
>> I re-booted.
>> BIG PROBLEMS!
>> I get 1:
>>        "modprobe: Can't open dependencies file
>>                /lib/modules/2.4.7-10/modules.dep not found."
>> and at least one:
>>        "Finding module dependencies: depmod: can't open
>>                /lib/modules/2.4.7-10/modules.dep for writing"
>> and 3 or 4:
>>        "modprobe: Can't open dependencies file
>>                /lib/modules/2.4.7-10/modules.dep
>>                     (No such file or directory.)"

> Big difficulties, since you apparently failed to bring the modules and
> rebuild your initrd if needed. Bare metal would be slower. Do this:

>     Boot the linux rescue CD.
>     Copy the kernel RPM over to "/tmp" on the mounted root partition disk
> (usually /mnt/sysimage).
>     chroot to the mounted disk (chroot /mnt/sysimage)
>     Install the RPM with "rpm -i /tmp/kernel*rpm"
>     Edit lilo.conf or grub.conf appropriately.
>     If you have a SCSI boot drive, you *MUST* do "mkinitrd" to generate a
> new set of pre-loadable SCSI modules that you have listed in lilo.conf or
> grub.conf
>     Run lilo or grub with the new kernel. You may have to edit some
> confusing and erroneous duplicate entries out of /etc/mtab in the mounted
> partition to make this work correctly.
>     Exit the chroot.
>     Reboot the machine.

This morning I (nervously) booted into linux single with the
rescue disk of my prior kernel version.  I first ran
/sbin/lilo -v -t which looked to be "ok".  
Then I ran /sbin/lilo for real.
Then I ran "init 6" to do a warm boot.
The system up and I am on the new kernel.

For all you lurkers out there:  In addition to the kernel HOWTO's,
you might want to review:
 www.redhat.com/docs/manuals/linux/RHL-7.2-Manual/custom-guide/rescuem...
Especially the part  a l l  t h e  w a y  towards the bottom
where it starts out "Have you ever rebuilt a kernel and, eager to try
out your new handiwork, rebooted before running /sbin/lilo?"

Again, thank you for having the patience and determination to
haunt these ngs and replying to rabble such as I.
I promise to do / to get better.
Jonesy
--
  | Marvin L Jones       | jonz           |  W3DHJ   |  OS/2

  |   7,703' -- 2,345m   |   frontier.net |  DM68mn             SK

For all you lurkers out there:  In addition to the kernel HOWTO's,
you might want to review:
 www.redhat.com/docs/manuals/linux/RHL-7.2-Manual/custom-guide/rescuem...
Especially the part  a l l  t h e  w a y  towards the bottom
where it starts out "Have you ever rebuilt a kernel and, eager to
try out your new handiwork, rebooted before running /sbin/lilo?"

Thank you for clearly and concisely pointing out my lapse.
Jonesy
--
  | Marvin L Jones       | jonz           |  W3DHJ   |  OS/2

  |   7,703' -- 2,345m   |   frontier.net |  DM68mn             SK

I must say I don't see the point of using RedHat's kernel upgrades.
If you want a new kernel why not just get the latest stable one
and compile it?
This is much easier than it sounds.

The trouble with RedHat kernels is that they have to cope with
every machine under the sun.
In particular they have to use initrd
in case the user has a SCSI machine.
(If you have SCSI discs you compile them into your own kernel,
and so don't need initrd.)

--
Timothy Murphy  

tel: 086-233 6090
s-mail: School of Mathematics, Trinity College, Dublin 2, Ireland

Quote:> Almost 2 weeks ago I did a RH 7.2 upgrade over my stable,
> fully functional RH 6.1.  I've been bleeding through one
> or both ears ever since.
[snip]
> Well, now I fear my system is TOTALLY hosed.
> I already had the new kernel -- downloaded by up2date
> last week, but not installed.
> I attempt to rpm -Uvh the kernel and I get "failed dependencies
>  modutils >= 2.4.10 is needed".

    NEVER UPGRADE YOUR KERNEL USING RPM.

That is not to say that you can't use the kernel RPMS. You should do

    rpm -ivh kernel-2.4.13*.rpm

which INSTALLs the new rpm.

This gives you both the new and your original kernels in the system. You should
check that the new kernel has been added to the lilo.conf file and that lilo
has been run. Now when you reboot the system, choose the new kernel from the
menu. If it fails, you can always go back to your older kernel.

Boot your damaged system with a bootdisc or repair CD. Install the older kernel
(you may have to do rpm -ivh kernel-2.4.whatever.rpm --force). Cross fingers
and reboot - pick the old kernel.

Cheers,
Toby Haynes

--

Toby Haynes
The views and opinions expressed in this message are my own, and do
not necessarily reflect those of IBM Canada.