[comp.unix.aix] Installed a buggy kernel ext. Duh, what do I do now?

dyer@spdcc.COM (Steve Dyer) (11/03/90)

I replaced /etc/nfs.ext with a test version of my own, ignorant
as I was of the baroque rebooting procedure of the RS6000 (not
to mention the bugs in my code!)

OK, it no longer boots, and flashes 888.  Hitting the clear button
once gives 120.  I assume it's crashing in the new module.

All I want to do right now is undo what I've done.  Isn't there
anyway to boot the damn thing single user before the rc scripts
start up, so I can rename the modules?

It isn't at all obvious to me why it was necessary for IBM to completely
redo everything so that the system administration knowledge gleaned by
users over the past 17 years becomes useless.

-- 
Steve Dyer
dyer@ursa-major.spdcc.com aka {ima,harvard,rayssd,linus,m2c}!spdcc!dyer
dyer@arktouros.mit.edu, dyer@hstbme.mit.edu

hbergh@oracle.nl (Herbert van den Bergh) (11/08/90)

In article <4754@spdcc.SPDCC.COM> dyer@ursa-major.spdcc.com (Steve Dyer) writes:
>I replaced /etc/nfs.ext with a test version of my own, ignorant
>as I was of the baroque rebooting procedure of the RS6000 (not
>to mention the bugs in my code!)

>All I want to do right now is undo what I've done.  Isn't there
>anyway to boot the damn thing single user before the rc scripts
>start up, so I can rename the modules?

I think it must be possible to reboot from floppies. I never tried it myself.


>It isn't at all obvious to me why it was necessary for IBM to completely
>redo everything so that the system administration knowledge gleaned by
>users over the past 17 years becomes useless.

Probably to give the users of the next 17 years a less hard time as they don't
need to memorize these configuration commands and files.

Could you go to some guy in your computer department who has been working with
*nix boxes for one month and tell him to extend the size of your /
filesystem with some megabytes, and expect the system to run? And I don't
mean DURING the job, because hardly any unix box can do that, but even 
after the job's done. You can show him how to do it 10 times and he`ll
screw up anyway.
What about extending the kernel (what you're working on right now, isn't it?).
How did you do that on other systems? Edit, compile, link kernel, reboot,
crash, reboot, edit, compile, link kernel, ...
With AIX 3.1 you can skip the 'link kernel, reboot' part, so that will save
enough time to study smit :-)

IBM, please don't ever port BSD or System V to these things.
Sell AIX to the others instead.

>-- 
>Steve Dyer
>dyer@ursa-major.spdcc.com aka {ima,harvard,rayssd,linus,m2c}!spdcc!dyer
>dyer@arktouros.mit.edu, dyer@hstbme.mit.edu

-- 
Herbert van den Bergh,		Email:	hbergh@oracle.nl, hbergh@oracle.com
ORACLE Europe

reilly@scotty.dccs.upenn.edu (G. Brendan Reilly) (11/09/90)

This is in response to the neophyte who wants IBM to port
SMIT to other systems:  Before this happens, make sure
that SMIT will give error messages when an error occurs.  I
just used SMIT to install 3002 and it decided to tell
everything was great even though the kernel was not updated.
Now, besides this, the csh it left on the system is invalid,
as well as other errors.  You may want to be careful about 3002.

When I called IBM Defect Support the person there asked me
to send a copy of the diskettes (all 16) to him since they
didn't have a copy as of yet.  I guess they got it later in
the day since the three other people I've spoken with
haven't asked me to send them their own software.

I stand with Steve - IBM made a major mistake redefining all
of the system management tools, and they are full of bugs.
The last thing any system needs is buggy system managment tools.

bengsig@oracle.nl (Bjorn Engsig) (11/09/90)

As a collegue of Herbert, I agree completely on his article
<1110@nlsun1.oracle.nl>.  To his note:
|Sell AIX to the others instead.
I just have the comment, that the first other system to have OSF/1 or AIX 3.1
should be the PS/2 - but you are working on that I guess.  Minidisks is really
a step backwards from the Logical Volume Manager.
-- 
Bjorn Engsig,         E-mail: bengsig@oracle.com, bengsig@oracle.nl
ORACLE Corporation    Path:   uunet!orcenl!bengsig

            "Stepping in others footsteps, doesn't bring you ahead"