SCIENTIFIC-LINUX-USERS Archives

November 2011

SCIENTIFIC-LINUX-USERS@LISTSERV.FNAL.GOV

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Yasha Karant <[log in to unmask]>
Reply To:
Yasha Karant <[log in to unmask]>
Date:
Tue, 15 Nov 2011 08:38:46 -0800
Content-Type:
text/plain
Parts/Attachments:
text/plain (75 lines)
On 11/15/2011 05:55 AM, Larry Linder wrote:
> On Monday 14 November 2011 11:41 pm, Yasha Karant wrote:
>> My wife did not notice that her laptop was unplugged, and ran the
>> machine until it quit for lack of power.  Running SL 6.1 IA-32, current
>> production SL kernel (not elrepo, etc.).  On the reboot with improperly
>> unmounted partitions, the auto fsck failed, and gave the usual enter
>> root password or ctrl-D prompt to run fsck manually.  When this
>> happened, rather than accept the full root password, after the first few
>> characters, the enter prompt re-appeared, and would never let me get to
>> a shell.  I finally had to put the install DVD that has the rescue
>> option into the DVD drive and boot from the DVD, and used fsck -y under
>> rescue to fsck the partitions from ls /dev/sd*.  The system then
>> rebooted.  What is wrong?  What do I need to change?
>>
>> Yasha Karant
>
> Reported this problem back in SL5.5 - it is still not fixed.   The only
> solution I found was to do an install and once disks were mounted - no
> formatting.   Then you do something crude - bail out at this point as new
> fsck has been written to match patricians.
> It works - desperate measures by desperate men.
> If only you were able to do what the message implies, have root passwd, edit
> required files, save files and reboot.
> There is a mathematical limit to security and that is when it becomes
> a "STONE".    It then only responds to a large hammer.
>
> One other issue is that during the install it knows all the disk info but
> there is no way to use it.   You have to plod threw the edit menu and
> manually fix stuff.
>
> Some one at RH need to put their foot down and say no more visual updates till
> core issues get fixed.   How do you like KDE and goofy icon stuff.   That is
> one reason we did not deploy SL 6.  we will stick with SL 5.7 for now.
>
> As a commercial user - I have a very different view of things.   If the
> factory is not running - nothing is shipped - no checks are received - no
> payroll money !  No wonder the Boss has but a little gray hair and is a
> grump.
>
> This is reality - like it or not.
>
> As a whole SL 5.7 is a great product, most love it, because it works - I only
> wish it could be better.
> These odd operational issues can be fixed and should be.
>
> Larry Linders

Although SL follows TUV, supposedly El Repo and some of the other 
repositories actually are willing to fix the bugs in the TUV 
distribution or to add vital functionality (such as additional device 
drivers for hardware not supported by TUV), perhaps one of these other 
EL compatible repositories will fix this issue?

The bug is somewhere in the software system that accepts keyboard input 
for the root password for manual intervention during boot.  Has the 
actual software causing the bug been identified?

Could the issue be SELinux?  This symptom does not appear on my 5.7 
laptop, but in general I force SELinux to use permissive mode. I have 
not tested this failure on a 6.1 system. I did not deactivate SELinux on 
my wife's laptop when I observed the boot fsck intervention failure.  I 
tried to set selinux=0 as a parameter to the kernel being booted in 
grub.conf as a workaround after the failure -- were SELinux to be the 
cause, but evidently this change would not be applied from the grub boot 
screen (grub active from the MBR, prior to loading any actual linux 
kernel) that purportedly permits editing.

As an aside following onto your "stone" comment, I do not recommend 
SELinux to most of my students for most situations; for 
professional/commercial use as a server, I recommend an appropriate 
(generally "high") throughput stateful inspecting dedicated hardware 
firewall defending the server -- unfortunately, an expensive solution.

Yasha Karant

ATOM RSS1 RSS2