Installation report , (ln, ls, semodule, rm)

[Jim Cornette]

Dawid Gajownik wrote:
> Hi!
>     I did today FTP network installation using boot.iso image from 9th 
> December. Almost everything went fine ;-) After reboot I saw kernel panic:
> 
> [snip]
> 
> Switching to new root and running init.
> unmounting old /dev
> unmounting old /proc
> unmounting old /sys
> Kernel panic - not syncing: Attempted to kill init!
> bla bla bla ;]
> 
> I looked at /root/install.log → http://pastebin.com/458435 and found 
> that there were problems with installing selinux-policy-targeted [1].
> 
> Instalowanie selinux-policy-targeted-2.1.1-1.noarch.
> /var/tmp/rpm-tmp.77057: line 22: ln: command not found
> /var/tmp/rpm-tmp.77057: line 34: semodule: command not found
> /var/tmp/rpm-tmp.77057: line 38: ls: command not found
> /var/tmp/rpm-tmp.77057: line 39: rm: command not found
> 
> I grabbed FC4 CD1, booted system into rescue mode and did `chroot 
> /mnt/sysimage'. I had to reinstall selinux-policy-targeted package with 
> `--noscript' option. I tried to manully execute all commands from 
> {pre,post}install scriplets but some of them failed (for example: 
> `semodule -b /usr/share/selinux/targeted/base.pp -s targeted'). I think 
> that the problem was with too old kernel available in FC4's rescue mode.
> 
> I also run `yum update' in chrooted environment but most of the 
> scriptlets failed. Unfortunately, it's not mentioned in /var/log/yum.log :/
> 
> After reboot 'kernel panic' problem disappeared :D There were lots of 
> avc messages, though. “Rescue mode” must have screw up security context 
> of system files. `touch /.autorelabel && reboot' fixed this problem...
> 
> Two remarks:
> 1) It would be nice to have newer rescue CD ;-)
> 2) I miss information about current download speed. For instance, 
> downloading stage2.img file takes some time on my DSL connection. I had 
> to look at NIC's LEDs to make shure that it's really downloading 
> something from the Internet.
> 
> That would be all. Now I can die happily ;-)
> 
> [1] BTW there are also problems with creating icon cache. Few scriplets 
> failed.
> 
> Regards,
>     Dawid
> 

You probably could have accessed your system by adding selinux=0 via 
grub on boot. Regardless, this problem seems to be related to Dec 2nd 
updates. SELinux policy seems to be effected directly or indrectly by a 
change somewhere.

I filed a bug that is now closed. The resolution you took resulted in 
similar results. Using --noscripts probably messed up something on the 
system though.

Bug hunters for similar failures should round up reports that are 
currently filed.

My bug report was filed after I resolved the issue. Another reporter is 
still running through the changes.

https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=174919

The bug above is for reference. You probably want anaconda involved with 
  you particular problem.

*Cannot link, run semodule, list or remove.*

You most likely hit the problem area closer to the cause.

Jim

-- 
"The best index to a person's character is a) how he treats people who 
can't
do him any good and b) how he treats people who can't fight back."
-- Abigail Van Buren