Subject: Re: Binary snapshot available
To: Jason Thorpe <thorpej@nas.nasa.gov>
From: Nathan Gelbard <gelbard@ENGR.ORST.EDU>
List: port-hp300
Date: 05/30/1997 13:49:51
At 01:01 PM 5/30/97 -0700, Jason Thorpe wrote:
> > Extracted usr.libexec. I think my 1st mistake was here.
> > 	I rebooted ;(
> > I got really familiar with the kernel debugger. 
>
>...what do you mean?  You rebooted... that should have worked ok... why
>did you need to use the debugger?  What happened, exactly?  Nothing
>should have gone wrong there.  If it did, there is a bug somewhere :-)

All of my daemons dumped core, and it went straight into the kernel
debugger. 

> > Extracted the rest, couldnt write over init, tar, gzip because
> > I was using them. I still havent found a way to overwrite them
> > because I need to run them to write over them! er...
>
>...sounds like you didn't follow the "special instructions" exactly.

whoops. forgot the --unlink. All of my robotic-shell scripts to
install new systems have that. Silly me ;0!

>mix them.  You should have made a backup of your ld.so first, like:
>
>	cp /usr/libexec/ld.so /usr/libexec/ld.so.bak

Did that.

>After extracting usr.libexec.tar.gz, you should have made a
>backup copy of your latest libc.so.*.  Then you should have extracted
>usr.lib.tar.gz.  Then the rest, in any order...

Did that too.

	(b) invoke tar like "tar --unlink -zxvpf /path/to/file.tar.gz"

Didnt scroll down enuf in my Xterm ;)

> > My scsi chasis/controller (?) is fubar, so I cant extract from my
> > 1.2 system.
>
>...did this happen only after the new kernel was booted?

No, on my OTHER system (hp370). I've always had problems with the
SCSI-controller & a group loop in the chasis.

> > fsck -p returns an error and messes up rc.
>
>What errors?  This information is important!

fsck returns a error to the shell. while -p is a valid option,
fsck wants more (like a filesystem to play with). the rc
script calls fsck to check all the filesytems, fsck returns
an error code (unknown) which gets caught by the case statement's
*), "Unknown error; help!"

> > telnetd no longer works.
>
>Define "no longer works"... are there errors?  Does it core?  Again,
>the details are important...

telnet connection refused from outside. Trying to telnet FROM the system (yes,
the network is configed) yeilds: arpresolve: can't allocate llinfo

> > Sigh. Looks like tomorrow its going to be a complete reinstall of 1.2.1.

Anyway I'm going to --unlink and pray ;)
Will post more after I've fixed/destroyed it!