Very Computer

I just 'completed' RH's Official Boxed (tm) upgrade to RH 7.2 on
my pre-loaded (i.e., I've barely messed with it) RH 6.1 system.

The upgrade went smoothly.  "Gee," I thought, "No bleeding
through the ears like on a OS/2 install or FixPack update."

Wrong!

On the re-boot I have no keyboard or mouse.
There's no kbd at the "press control-X for text mode" screen.

I tried booting "linux rescue" with the boot disk created as
the last step in the RH 7.2 upgrade process.  Same results.

I _can_ boot "linux rescue" with the boot.img I was smart/lucky
enough to put on a floppy before I ever started into this.

Be gentle: I'm a Real Rookie with linux.  I can noodle around
on the /mnt/sysimage that "linux rescue" gives me  --  but,
I have no idea what config files I need to futz with - where
they'd be located - and, what the contents ought to be.

The keyboard is unremarkable (it does have the MicroSloth keys),
and the mouse, too, is unremarkable: PS2.  Both worked in 6.1.

thing re-booted with a keyboard.  Working in "rescue" is No Fun.

Many thanks in advance,
Jonesy
--
  | Marvin L Jones       | jonz           |  W3DHJ   |  OS/2

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

Quote:> I just 'completed' RH's Official Boxed (tm) upgrade to RH 7.2 on
> my pre-loaded (i.e., I've barely messed with it) RH 6.1 system.

> The upgrade went smoothly.  "Gee," I thought, "No bleeding
> through the ears like on a OS/2 install or FixPack update."

> Wrong!

> On the re-boot I have no keyboard or mouse.
> There's no kbd at the "press control-X for text mode" screen.

And it's not a USB keyboard. Gack. Can you boot from the installation CD
with "linux rescue", and do a comparison of the installed RPM's to the new
RH 7.2 ones on the CD? Or use the CD, chroot to the mounted root partition,
and start up the network to grab any interesting kbd or mouse upgrades, or
the kernel update? And can you call RedHat for support?

Quote:> I tried booting "linux rescue" with the boot disk created as
> the last step in the RH 7.2 upgrade process.  Same results.

That's *. Sounds like a kernel/hardware compatibility issue.

Quote:> I _can_ boot "linux rescue" with the boot.img I was smart/lucky
> enough to put on a floppy before I ever started into this.

Cool.

We need the hardware names and models, and especially the system type and
BIOS version.

There's no way of not having a keyboard or mouse. It's a kernel driver
that has to be compiled IN (psaux). He should boot in single user mode
(or even with init=/bin/sh - remember to mount -wno remount /, and sync,
sync, sync before rebooting) in order to stop the standard bootup
scripts running in order to confirm that the raw kernel leaves him with
mouse and keyboard access.

Quote:> RH 7.2 ones on the CD? Or use the CD, chroot to the mounted root partition,
> and start up the network to grab any interesting kbd or mouse upgrades, or
> the kernel update? And can you call RedHat for support?

He heh.

Quote:>> I tried booting "linux rescue" with the boot disk created as
>> the last step in the RH 7.2 upgrade process.  Same results.
> That's *. Sounds like a kernel/hardware compatibility issue.

It does, but he needs to confirm that, via the trick I mentioned.
He might have to recompile the kernel for his architecture, if the
kernel he is booting i snot a vanilla 386 kernel. There ARE also
some tricks about enabling the keyboard gate that might need
checking for his hardware.

In any case, he should boot with his old 6.2 kernel and that'll be
that.

Quote:>> I _can_ boot "linux rescue" with the boot.img I was smart/lucky
>> enough to put on a floppy before I ever started into this.

Eh? You mean that's the old kernel? Good. Use it.

Quote:> We need the hardware names and models, and especially the system type and
> BIOS version.

Also very good idea. But he can just stuff his working boot kernel on
disk and use it from now on until he learns better.

Peter

Excuse me, but you haven't some of the weird all locally compiled kernels I
have, and some of the driver/hardware incompatibilities. Also, re-read his
note: when he is booting from the MBR and getting the "select" screen for
LILO or grub (whichever he is using), he doesn't have a keyboard to work
with. His only hope to do the kind of manipulations you describe is to let
it boot, log in remotely, and set the boot status by hand before rebooting.

Quote:> > RH 7.2 ones on the CD? Or use the CD, chroot to the mounted root
partition,
> > and start up the network to grab any interesting kbd or mouse upgrades,
or
> > the kernel update? And can you call RedHat for support?

> He heh.

> >> I tried booting "linux rescue" with the boot disk created as
> >> the last step in the RH 7.2 upgrade process.  Same results.

> > That's *. Sounds like a kernel/hardware compatibility issue.

> It does, but he needs to confirm that, via the trick I mentioned.
> He might have to recompile the kernel for his architecture, if the
> kernel he is booting i snot a vanilla 386 kernel. There ARE also
> some tricks about enabling the keyboard gate that might need
> checking for his hardware.

He can't get that far, remember? No keyboard at boot time after the MBR
loads and displays the kernel selections.

Quote:> In any case, he should boot with his old 6.2 kernel and that'll be
> that.

You are making a *huge* and probably incorrect assumption: if he used the
RedHat CD "upgrade" technique, a huge set of previous RPM's, including
kernels, got replaced.

Quote:> >> I _can_ boot "linux rescue" with the boot.img I was smart/lucky
> >> enough to put on a floppy before I ever started into this.

> Eh? You mean that's the old kernel? Good. Use it.

That is a good hint.

Quote:> > We need the hardware names and models, and especially the system type
and
> > BIOS version.

> Also very good idea. But he can just stuff his working boot kernel on
> disk and use it from now on until he learns better.

> Peter

It's a start. But it's going to potentially yack up hairballs if he's using
ext3: he'll need to edit his /etc/fstab if so, to restore ext2 usage.

And some things I forgot: if he's going from his RH 7.2 installation to the
old RH 6.1 kernel on his floppy, then he'll need to duplicate the contents
of "/etc/modules.conf" to "/etc/conf.modules", at least if his disk is on a
SCSI controller.

Switching critical components of your OS to ones from a previous release is
often under-estimated in difficulty: there are lots of dependencies, not
just the ones listed in the RPM's ".spec" file, and it's impossible to put
in an older RPM all the conflicts with newer packages because they didn't
exist at the time the RPM was made and were thus impossible to predict.

 [ I'm going to followup here, at this point in the thread.  I _have_ ]
 [ read the followon postings.  But, posting here is "simpler".       ]

On Sat, 9 Feb 2002 12:45:01 +0100, Peter T. Breuer scribbled:

Well, you're correct, and I am Rong!  As I lay tossing and turning
in bed last night with infuriation -- and with my wife -- I got a
wee little thought:  The kbd works up through the massive boot
scrolling, because after the "Welcome To Red Hat linux" + "Press 'I'
to enter interactive startup" text scrolls up, I can use ^S and ^Q
to control the scroll of the messages.  So, this morning I fired it
up and when it through up the login window (with the dead kbd and
mouse) I ctl-alt-BACKSPACE'd the sucker.  That worked!

Big deal -- it just re-cycled and threw up another login window.
"Groan!", I thought.
BUT, THIS TIME THE KEYBOARD AND MOUSE WORKED!!!

Quote:>>> I _can_ boot "linux rescue" with the boot.img I was smart/lucky
>>> enough to put on a floppy before I ever started into this.

> Eh? You mean that's the old kernel? Good. Use it.

No.  This one was/is the boot.img from the RH 7.2 CD/images dir.
Of course, it never gets anywhere near XWindows.

I've re-booted the thing several times now.  The ctl-alt-BKSP always
'does it' for me.  Except - one time - I can swear it booted without
the lockup in XWindows.

So now I have at least a well-enough-functioning system that I can
probably all fix the 'things' (such as this snafu XWindows) that are
awry.  For one thing, about once per every 4-5 boots, kudzu pops
up and annouces it is removing my mouse.  kudzu also almost always
removes my sound -- and sound is _not_ useable when it _does_ get
installed.

But, that's grist for other threads -- and, may well be covered
already.  I'll go google up the symptoms.

As for XWindow:  Any ideas?  I've yet to go to reahat.com and see
if there are any updates/fixes for it.  I'll do that -- but I wanted
to reply to the folks here that were nice enough to work with me.

Thanks for your replies.  It sure helps when you're all alone
with problems such as these.

Regards,
Jonesy
--
  | Marvin L Jones       | jonz           |  W3DHJ   |  OS/2

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

Oh - is that what he means by "press ctl-X ..."? I've never seen such a
screen. I thought he was referring to some redhat install or postinstall
routine.

Well, if he doesn't have a keyboard or mouse when lilo is running, then
his machine is not accepting inb/outb stuff to the low memory addresses
where they should be. Therefore his machine is not doing some bios
setup stuff that it should do.

His best bet is to wait for a timeout. The ps ports might get
initialized when the kernel does its Setup.S stuff.

Quote:> with. His only hope to do the kind of manipulations you describe is to let
> it boot, log in remotely, and set the boot status by hand before rebooting.

"Let it boot" I think is the key! With luck, that'll be enough.

Quote:>> > RH 7.2 ones on the CD? Or use the CD, chroot to the mounted root partition,
>> > and start up the network to grab any interesting kbd or mouse upgrades, or
>> > the kernel update? And can you call RedHat for support?

>> He heh.

>> >> I tried booting "linux rescue" with the boot disk created as
>> >> the last step in the RH 7.2 upgrade process.  Same results.

>> > That's *. Sounds like a kernel/hardware compatibility issue.

>> It does, but he needs to confirm that, via the trick I mentioned.
>> He might have to recompile the kernel for his architecture, if the
>> kernel he is booting is not a vanilla 386 kernel. There ARE also
>> some tricks about enabling the keyboard gate that might need
>> checking for his hardware.
> He can't get that far, remember? No keyboard at boot time after the MBR
> loads and displays the kernel selections.

I suspect it might come back later on in the boot.

But I like your suggestion of logging in remotely and fixing things
from "there".

Quote:>> In any case, he should boot with his old 6.2 kernel and that'll be
>> that.
> You are making a *huge* and probably incorrect assumption: if he used the
> RedHat CD "upgrade" technique, a huge set of previous RPM's, including
> kernels, got replaced.

Nevertheless, he can put his old modules back, if they got vamooshed
(thyey'll be in some rpm on a cd) And that should be enough, with the
old kernel.

Quote:>> >> I _can_ boot "linux rescue" with the boot.img I was smart/lucky
>> >> enough to put on a floppy before I ever started into this.

>> Eh? You mean that's the old kernel? Good. Use it.
> That is a good hint.
>> > We need the hardware names and models, and especially the system type and
>> > BIOS version.

>> Also very good idea. But he can just stuff his working boot kernel on
>> disk and use it from now on until he learns better.
> It's a start. But it's going to potentially yack up hairballs if he's using
> ext3: he'll need to edit his /etc/fstab if so, to restore ext2 usage.

Doesn't redhat engineer for that? I always add ext3 partitions as type
"auto", and make mount.auto read /etc/filesystems to determine in which
order it should try. And I make the order go ext3 ext2 ....

Umm, if his root filesystem is type ext3, and his kernel has support
only for type ext2, then I think he's OK, as it will mount as type
ext2, surely? Modulo the kernel 2.4.17 root-ext3 problem! Well, he's
probably the first person to try it. Yes, he probably may need to
run a live cd system and fix his fstab back to "appropriate levels of
failsafety".

Peter

Oh. You mean everything works up until you get an X login screen (xdm).

OK, fine. Don't start in X. Go to a login prompt instead (i.e start in
run level 3, or thereabouts). Then do a "startx".

If that works, we'll take it from there (reasoning: xdm has a bad
interaction with some things).

The "from there" includes such banalities as "you have misconfigured
the mouse", and "stop running gpm".

Quote:> Big deal -- it just re-cycled and threw up another login window.
> "Groan!", I thought.
> BUT, THIS TIME THE KEYBOARD AND MOUSE WORKED!!!

OK. Work it out later. Sounds like a networking timeout. Make sure
your machine is listed in /etc/hosts (and that localhost is also
listed, etc, etc, - redhat classically fouls up /etc/hosts something
awful).

Peter

On Sat, 9 Feb 2002 23:35:48 +0100, Peter T. Breuer scribbled:

<- grosse geschnip ->

Quote:>> The kbd works up through the massive boot
>> scrolling, because after the "Welcome To Red Hat linux" + "Press 'I'
>> to enter interactive startup" text scrolls up, I can use ^S and ^Q
>> to control the scroll of the messages.  So, this morning I fired it
>> up and when it through up the login window (with the dead kbd and
>> mouse) I ctl-alt-BACKSPACE'd the sucker.  That worked!

> Oh. You mean everything works up until you get an X login screen (xdm).

yep.  And, being a rookie in linux, I did not understand the 'subtlety'.

Quote:> OK, fine. Don't start in X. Go to a login prompt instead (i.e start in
> run level 3, or thereabouts). Then do a "startx".

Well, ctl-alt-BKSP 'works' at this point.  And, isn't it just
doing (another) "startx" to throw up the login panel again?

Quote:>> Big deal -- it just re-cycled and threw up another login window.
>> "Groan!", I thought.
>> BUT, THIS TIME THE KEYBOARD AND MOUSE WORKED!!!

> OK. Work it out later. Sounds like a networking timeout. Make sure
> your machine is listed in /etc/hosts (and that localhost is also
> listed, etc, etc, - redhat classically fouls up /etc/hosts something
> awful).

hmmmm...  Would that be 'perturbed' by not having all the "hosts"
on the lan powered up?  I have 2 OS/2 machines and a (BLEECH!) Win95
machine on the lan.  The Win95 was _not_ powered up during any
of this 'activity'.  
I'll test it tomorrow with and without all the machines powered up
and running.

(It looks like /etc/hosts was good-to-go -- all the other (3) machines
on the (OS/2 Peer-To-Peer) lan were in there, along with "localhost".  
I've since renamed "localhost" to a more better nom-de-network.

The RH 7.2 upgrade is bumbling along reasonably well.  There's a lot
to be corrected/re-worked/fixed.  The original system was a pre-load
that included 'features' of both a workstation and a server.  It
looks like the upgrade screwed up some of the sever stuff:  I can
http into the box -- but telnet and ftp are still not working.

Again, thnks for your time and thoughts.
Jonesy
--
  | Marvin L Jones       | jonz           |  W3DHJ   |  OS/2

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

  [[ I see my ISP's newsserver didn't get this out the 1st time. ]]

On Sat, 9 Feb 2002 23:35:48 +0100, Peter T. Breuer scribbled:

<- grosse geschnip ->

Quote:>> The kbd works up through the massive boot
>> scrolling, because after the "Welcome To Red Hat linux" + "Press 'I'
>> to enter interactive startup" text scrolls up, I can use ^S and ^Q
>> to control the scroll of the messages.  So, this morning I fired it
>> up and when it through up the login window (with the dead kbd and
>> mouse) I ctl-alt-BACKSPACE'd the sucker.  That worked!

> Oh. You mean everything works up until you get an X login screen (xdm).

yep.  And, being a rookie in linux, I did not understand the 'subtlety'.

Quote:> OK, fine. Don't start in X. Go to a login prompt instead (i.e start in
> run level 3, or thereabouts). Then do a "startx".

Well, ctl-alt-BKSP 'works' at this point.  And, isn't it just
doing (another) "startx" to throw up the login panel again?

Quote:>> Big deal -- it just re-cycled and threw up another login window.
>> "Groan!", I thought.
>> BUT, THIS TIME THE KEYBOARD AND MOUSE WORKED!!!

> OK. Work it out later. Sounds like a networking timeout. Make sure
> your machine is listed in /etc/hosts (and that localhost is also
> listed, etc, etc, - redhat classically fouls up /etc/hosts something
> awful).

hmmmm...  Would that be 'perturbed' by not having all the "hosts"
on the lan powered up?  I have 2 OS/2 machines and a (BLEECH!) Win95
machine on the lan.  The Win95 was _not_ powered up during any
of this 'activity'.
I'll test it tomorrow with and without all the machines powered up
and running.

(It looks like /etc/hosts was good-to-go -- all the other (3) machines
on the (OS/2 Peer-To-Peer) lan were in there, along with "localhost".
I've since renamed "localhost" to a more better nom-de-network.

The RH 7.2 upgrade is bumbling along reasonably well.  There's a lot
to be corrected/re-worked/fixed.  The original system was a pre-load
that included 'features' of both a workstation and a server.  It
looks like the upgrade screwed up some of the sever stuff:  I can
http into the box -- but telnet and ftp are still not working.
In fact, it looks like telnetd and ftpd are flat out gone now.  sigh...

Again, thanks for your time and thoughts.
 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

Much Later (so much so I noticed my first post above did not 'stick'.):

After about 5-6 re-boots today, my ctl-alt-BKSP 'trick' no longer
works.  I end up at the login window -- the kbd and mouse are dead --
and only a hardware reset will 'work'.

I see my big, blue box with OS/2 Warp 4 on the shelf across the
room.  It seems to be beckoning to me...

Jonesy
--
  | Marvin L Jones       | jonz           |  W3DHJ   |  OS/2

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

Quote:> The RH 7.2 upgrade is bumbling along reasonably well.  There's a lot
> to be corrected/re-worked/fixed.  The original system was a pre-load
> that included 'features' of both a workstation and a server.  It
> looks like the upgrade screwed up some of the sever stuff:  I can
> http into the box -- but telnet and ftp are still not working.

> Again, thnks for your time and thoughts.
> Jonesy

They're disabled by default, and in most cases shouldn't even be installed
(in my opinion). If you need that sort of access, install and use the
OpenSSH server.

No - it's your OWN hostname that must be in there. If you are missing
that entry, then, yes, connection with your own machine from itself may
depend on other machines on the net in all sorts of ways, but the
brokenness is in your machine's misconfiguraton.

Quote:> machine on the lan.  The Win95 was _not_ powered up during any
> of this 'activity'.  
> (It looks like /etc/hosts was good-to-go -- all the other (3) machines
> on the (OS/2 Peer-To-Peer) lan were in there, along with "localhost".  

That's wrong! Localhost MUST be there. It's the name for the local
machine! It is not a changable thing!

Quote:> I've since renamed "localhost" to a more better nom-de-network.

No no no noo.

The correct entries are:

127.0.0.1 localhost loghost
192.168.1.3 me.mine.org me

where "me" is your machine, and 192.168.1.3 is the address of its
ethernet card.

Peter

On Sun, 10 Feb 2002 14:19:17 +0100, Peter T. Breuer scribbled:

This is worrisome...

Quote:>> I've since renamed "localhost" to a more better nom-de-network.

> No no no noo.

I did not do it by hacking in the hosts file directly.
I used the GUI (bleech) config program to rename "localhost"
to "linus".  (I know -- but, it was my wife's idea and how many
PC's on a lan can you have called "localhost"?)

Quote:> The correct entries are:

> 127.0.0.1 localhost loghost
> 192.168.1.3 me.mine.org me

> where "me" is your machine, and 192.168.1.3 is the address of its
> ethernet card.

I haven't yet booted to the "linux rescue" mode yet this morning.
But, my host file has _no_ 'alpha' domain names in it.  It just
has the IP addresses of the machines on the lan, and their 'host'
name.  ('host' - that is such a poor choice of words, IMO. There
are no 'hosts' on a lan -- It's more like a "no host party".  
i.e.: anarchy. :-)

From memory:
192.168.0.5  linus
192.168.0.3  green
192.168.0.4  blue
192.168.0.2  red   # with 'descriptions' as comment in all 4 entries

(I know -- the other names are pretty lame, too. )
( .0.1 is/was a retired 486 -- our first OS/2 machine.)

Now to try the first boot of the day....

Jonesy
--
  | Marvin L Jones       | jonz           |  W3DHJ   |  OS/2

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

On Sun, 10 Feb 2002 06:54:55 GMT, Nico Kadel-Garcia scribbled:

I understand.  But, they were there, and being used in the (working)
RH 6.1 install.  I take exception to the fact that the RH 7.2 _upgrade_
saw fit to delete telnetd and ftpd.  Elsewise, I could telenet into
the thing from my OS/2 workstation and do the re-work _much_ easier.

I'll move on to SSH once I get the linux machine back from it's coma.

Jonesy
--
  | Marvin L Jones       | jonz           |  W3DHJ   |  OS/2

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

All of them. That's the point. It means "I". Everyone in the universe
is called "I".

Quote:>> The correct entries are:

>> 127.0.0.1 localhost loghost
>> 192.168.1.3 me.mine.org me

>> where "me" is your machine, and 192.168.1.3 is the address of its
>> ethernet card.
> I haven't yet booted to the "linux rescue" mode yet this morning.
> But, my host file has _no_ 'alpha' domain names in it.  It just
> has the IP addresses of the machines on the lan, and their 'host'

This is incorrect. The FQDN of the machine must be the first entry
after the IP address on the line, and then it is followed by its
aliases, which may or may not be short forms of the FQDN.

Quote:> name.  ('host' - that is such a poor choice of words, IMO. There
> are no 'hosts' on a lan -- It's more like a "no host party".  
> i.e.: anarchy. :-)
> From memory:
> 192.168.0.5  linus

You need a FQDN. Without it, many things will break. Mail for one
thing.

Quote:> 192.168.0.3  green
> 192.168.0.4  blue
> 192.168.0.2  red   # with 'descriptions' as comment in all 4 entries
> (I know -- the other names are pretty lame, too. )
> ( .0.1 is/was a retired 486 -- our first OS/2 machine.)
> Now to try the first boot of the day....

Remember that the entry

  127.0.0.1 localhost loghost

is _required_. It is the loopback interface.

Peter