How to capture "make installworld" error during migration from 5.1 to 5-Stable?

[Christian Hiris]

On Wednesday 01 December 2004 07:05, Stacey Roberts wrote:
> Hello Christain,
>
> ----- Original Message -----
> From: "Christian Hiris <4711 at chello.at>"
> To: To freebsd-questions at freebsd.org
> Date: Wed, 01 Dec, 2004 05:40 GMT
> Subject: Re: How to capture "make installworld" error during migration from
> 5.1 to 5-Stable?
>
> > -----BEGIN PGP SIGNED MESSAGE-----
> > Hash: SHA1
> >
> > On Wednesday 01 December 2004 02:12, Stacey Roberts wrote:
> > > Okay.., I've just scripted the output from "make installworld" at step
> > > 15 of the migration guide, and its failed as before.
> > >
> > > I've got the output redirected to a file: /var/tmp/installworld.log,
> > > however, I'm in single-user mode, and nothing works: ls, mount, etc..,
> > > how do I get this file off this system so that I can post it to the
> > > list for assistance?
> > >
> > > At this point, I can (manually) show the last bit of the failed attempt
> > > for make installworld:
> > >
> > > ===> bin/test
> > > install -s -o root -g wheel -m 555  test /bin
> > > install -o root -g wheel -m 444 test.1.gz /usr/share/man/man1
> > > pid 45090 (sh), uid 0: exited on signal 10 (core dumped)
> > > *** Signal 10
> > > Stop in /usr/src/bin/test.
> > > *** Error code 1
> > > Stop in /usr/src.
> > > *** Error code 1
>
> I've managed to get the log file off the machine. I've attached it here in
> gzipped format for anyone that is able (including yourself, if you like) to
> be able to look at it in its entirety.
>
> > This eventually points out a problem in your memory or harddisk
> > subsystem, I had such errors on incompatible disk controllers starting at
> > 5.1. This kind of errors silently destroyed the data on my system
> > (happened with HPT onboard controllers on an Via-KT600 board). Double,
> > better, triple fsck your filesystems and check CFLAGS and COPTFLAGS
> > settings in /etc/make.conf.
>
> I've found that I can reboot the system into multiuser mode, however there
> *are* lots of programs core-dumping all over the place (sendmail, exited on
> signal 11 - for instance)..,

This happens because you install target fails and there are a lot of old 
binaries installed on your system. 

>
> > Did you set your kernel timezone with "adjkerntz -i"?
>
> Yes.., followed the migration guide to the letter, save for scripting step
> 15 as I originally asked about here..,
>
> I'm actually preparing to head off-site here at present, but will get back
> to this later on today.., hopefully there'd be more information on the
> situation after examination of the log file output by a kind soul..,

The only thing I can find, is that the install of the test man page fails. I 
can't tell you why this happens. Maybe there is a problem with removing your 
old test.1.gz (ie. disk access error). 
First I would try to manually remove the test man-page by the command  
"/bin/rm  /usr/share/man/man1/test.1.gz" and run the install target again.

If the install target still fails with test.1.gz, have a closer look 
on /usr/obj/usr/src/bin/test/test.1.gz. I attached 5.3 version of it as of 
Nov, 26th.

# cd /usr/obj/usr/src/bin/test
# ls -l
total 24
-rw-r--r--  1 root  wheel  1800 Nov 26 16:14 .depend
-rwxr-xr-x  1 root  wheel  9839 Nov 26 16:21 test
-rw-r--r--  1 root  wheel  2654 Nov 26 16:21 test.1.gz
-rw-r--r--  1 root  wheel  6708 Nov 26 16:21 test.o
#

-- 
Christian Hiris <4711 at chello.at> | OpenPGP KeyID 0x3BCA53BE 
OpenPGP-Key at hkp://wwwkeys.eu.pgp.net and http://pgp.mit.edu