Can't start Mandriva

On Sun, 08 Apr 2007 11:45:34 -0400, David W. Hodgins wrote:
[color=blue]
> Yes. Also check /var/log/syslog, after a failed startx, for any
> related error messages. /var/log/messages, /var/log/kernel/errors too.[/color]

Jeebus! Those files are ginormous!

They point at a restart event today at 12.01.57 and although my 8 year old
lad wouldn't admit to it, I suspect he hit the reset button on the front
of the tower, not realising that the PC was running as the monitor was
switched off at the time. There are no error messages in
/var/log/kernel/errors at all. A blank page.

I suppose I'll wade through reams of messages in the files you mentioned
to see if I can come across any glaringly obvious fubars.

What's hard to understand is that, according to the logs, it was running
well right up to the moment it was restarted and yet it went on the fritz
as soon as it got as far as X.

Is it possible to do an upgrade install that will right any wrongs in some
conf file somewhere? It would save me looking for this needle in a
haystack.

On 2007-04-08, Aardvark <aardvark@youllnever.know> wrote:[color=blue]
>
> Any help gladly accepted.[/color]

These kind of things Ive found are solve through log files and
elimination :) HTH :):)

What happens if you log in under a different account? A new account?
User a different window manager?
Somewhere - hopefully - is a log file with your error in it :(

Lordy[color=blue]
>[/color]

On Sun, 08 Apr 2007 11:45:34 -0400, David W. Hodgins wrote:
[color=blue]
> Yes. Also check /var/log/syslog, after a failed startx, for any
> related error messages. /var/log/messages, /var/log/kernel/errors too.[/color]

Dunno whether this means anything or not, but the following two lines
appear repeatedly in both syslog and messages shortly after the trouble
began:

Apr 8 12:05:49 aardvark2 cfservd[3600]: Input file /etc/cfengine/cfservd.conf missing or busy..
Apr 8 12:05:49 aardvark2 cfservd[3600]: /etc/cfengine/cfservd.conf: No
such file or directory

Aardvark <aardvark@youllnever.know> writes:
[color=blue]
>On Sun, 08 Apr 2007 11:45:34 -0400, David W. Hodgins wrote:[/color]
[color=blue][color=green]
>> Yes. Also check /var/log/syslog, after a failed startx, for any
>> related error messages. /var/log/messages, /var/log/kernel/errors too.[/color][/color]
[color=blue]
>Dunno whether this means anything or not, but the following two lines
>appear repeatedly in both syslog and messages shortly after the trouble
>began:[/color]
[color=blue]
>Apr 8 12:05:49 aardvark2 cfservd[3600]: Input file /etc/cfengine/cfservd.conf missing or busy..
>Apr 8 12:05:49 aardvark2 cfservd[3600]: /etc/cfengine/cfservd.conf: No
>such file or directory[/color]

rpm -e cfengine-base cfengine-cfservd

No idea why you have it installed.
cfengine is a language based system specifically designed for testing
and configuring unix-like systems attached to a TCP/IP network.

On Sun, 08 Apr 2007 21:16:24 +0000, Unruh wrote:
[color=blue]
>
> No idea why you have it installed.
> cfengine is a language based system specifically designed for testing
> and configuring unix-like systems attached to a TCP/IP network.[/color]

LOL Well that'll be going when I get back into Mandriva :-)

Just came a recent bug report to Mandriva. The symptoms appear to be
similar to what I've been getting.

[url]http://archives.mandrivalinux.com/cooker/2007-03/msg00473.php[/url]

Hmmmmm.

On Sun, 08 Apr 2007 19:25:49 +0000, Aardvark wrote:
[color=blue]
> On Sun, 08 Apr 2007 11:45:34 -0400, David W. Hodgins wrote:
>[color=green]
>> Yes. Also check /var/log/syslog, after a failed startx, for any
>> related error messages. /var/log/messages, /var/log/kernel/errors too.[/color]
>
> Jeebus! Those files are ginormous!
>
> They point at a restart event today at 12.01.57 and although my 8 year old
> lad wouldn't admit to it, I suspect he hit the reset button on the front
> of the tower, not realising that the PC was running as the monitor was
> switched off at the time. There are no error messages in
> /var/log/kernel/errors at all. A blank page.
>
> I suppose I'll wade through reams of messages in the files you mentioned
> to see if I can come across any glaringly obvious fubars.
>
> What's hard to understand is that, according to the logs, it was running
> well right up to the moment it was restarted and yet it went on the fritz
> as soon as it got as far as X.
>
> Is it possible to do an upgrade install that will right any wrongs in some
> conf file somewhere? It would save me looking for this needle in a
> haystack.[/color]

IF..... you were usig 3D-desktop.

At login select a non 3D desktop and follow instructions.

There is/was a bug which will not allow a new login using 3D.

Aardvark <aardvark@youllnever.know> writes:
[color=blue]
>Just came a recent bug report to Mandriva. The symptoms appear to be
>similar to what I've been getting.[/color]
[color=blue]
>[url]http://archives.mandrivalinux.com/cooker/2007-03/msg00473.php[/url][/color]

Sorry, are you running cooker? If so, you are on your own. cooker is
supposed to have broken stuff in it, so your finding some is just what
anyone would expect.

On Sun, 08 Apr 2007 19:09:39 -0400, Aardvark <aardvark@youllnever.know> wrote:
[color=blue]
> Just came a recent bug report to Mandriva. The symptoms appear to be
> similar to what I've been getting.
>
> [url]http://archives.mandrivalinux.com/cooker/2007-03/msg00473.php[/url][/color]

Not the same. The bug in the cooker version is affecting xdm,
not startx. xdm, or kdm, or gdm was working on your system,
but failing when you selected to start kde/gnome, whichever.
Which dm are you using? Is your default startup kde?

Also, it's normal to see messages from various applications
about missing ipv6, when ipv6 is disabled (as it noramlly should
be).

You mentioned that your log files were very large. Do you have
anacron installed, so the log rotation script will run each Sunday?
By default, it's run a 4 am, so unless you leave your computer on
overnight, it won't run without anacron.

When you're viewing syslog, go to the end, and then search back for
"syslogd", or "restart", to find the latest boot.

Regards, Dave Hodgins

--
Change nomail.afraid.org to ody.ca to reply by email.
(nomail.afraid.org has been set up specifically for
use in usenet. Feel free to use it yourself.)

On Mon, 09 Apr 2007 08:54:24 -0400, David W. Hodgins wrote:
[color=blue]
> On Sun, 08 Apr 2007 19:09:39 -0400, Aardvark <aardvark@youllnever.know> wrote:
>[color=green]
>> Just came a recent bug report to Mandriva. The symptoms appear to be
>> similar to what I've been getting.
>>
>> [url]http://archives.mandrivalinux.com/cooker/2007-03/msg00473.php[/url][/color]
>
> Not the same. The bug in the cooker version is affecting xdm,
> not startx. xdm, or kdm, or gdm was working on your system,
> but failing when you selected to start kde/gnome, whichever.
> Which dm are you using? Is your default startup kde?
>[/color]

Yup. KDE 3.5.something. I didn't really think I had the Cooker bug- it
just struck me how similar the symptoms were.
[color=blue]
> Also, it's normal to see messages from various applications
> about missing ipv6, when ipv6 is disabled (as it normally should
> be).
>[/color]

I had read that somewhere. A second opinion never goes amiss though :-)
[color=blue]
> You mentioned that your log files were very large. Do you have
> anacron installed, so the log rotation script will run each Sunday?
> By default, it's run a 4 am, so unless you leave your computer on
> overnight, it won't run without anacron.
>[/color]

When I installed Mandriva I set the logs to be as verbose as possible
thinking that if anything goes wrong you can never get too much
information. All the logs did start at shortly after 4.am Sunday (the PC
usually runs all night as I leave downloads running overnight) although I
did only examine the section from just after 12 noon which was when the
problem first arose.
[color=blue]
> When you're viewing syslog, go to the end, and then search back for
> "syslogd", or "restart", to find the latest boot.
>[/color]

That's what I did. If memory serves me right the restart was at 12.01.48
and the previous log entries were just before 12. When I get some time
today I think I'll edit the Mandriva GRUB file to run the old kernel
(2.6.15-5) to check if my problem is specific to the new kernel or happens
regardless.

Do you know what the most recent kernel version is? I might try that when
I get Mdv running.
[color=blue]
> Regards, Dave Hodgins[/color]

Cheers, Steve Doherty

On Mon, 09 Apr 2007 11:39:43 -0400, Aardvark <aardvark@youllnever.know> wrote:
[color=blue]
> When I installed Mandriva I set the logs to be as verbose as possible
> thinking that if anything goes wrong you can never get too much
> information. All the logs did start at shortly after 4.am Sunday (the PC
> usually runs all night as I leave downloads running overnight) although I
> did only examine the section from just after 12 noon which was when the
> problem first arose.[/color]

Heh, heh. As you've learned, the hard part of using the log files to
track down the cause of a problem, is figuring out which entries actually
relate to the problem. That's a skill that develops over time.
[color=blue]
> Do you know what the most recent kernel version is? I might try that when
> I get Mdv running.[/color]

After running urpmi.update, I use "urpmq kernel-source" to see which is
the latest available, on the mirrors I use. There may be later versions,
for which the source hasn't propagated yet, but those are of no use to
me, as I have to recompile my winmodem driver, after each kernel install.

At present, the latest is ...
$ urpmq kernel-source
no package named kernel-source
The following packages contain kernel-source:
kernel-source-2.6.17.13mdv
$ uname -r
2.6.17-13mdv

Pay attention on the first boot after installing a new kernel, ensuring
the video drivers are installed/selected properly.

Also, if you had a hard restart (whatever the cause<G>), I would boot
from a cd, or other running linux system, and fsck all of the filesystems
related to the failing install.

If you're using an ext filesystem, check the lost+found, to see if there
are any obvious files that have been lost.

I highly recommend turning off hard drive write caching, as the small
performance gain, is greatly outweighed by the potential loss caused
by a hard reset. I learned this the hard way. Even with a journalized
file system, if the journal entries haven't been physically written
to disk, when the kernel module assumes they have, you can lose entire
file systems.

I have the following in /etc/rc.d/rc.local ...
hdparm -a 8 -d 1 -r 0 -k 1 -u 1 -m 16 -c 1 -A 1 -W 0 -S 240 /dev/hda
hdparm -a 8 -d 1 -r 0 -k 1 -u 1 -m 16 -c 1 -A 1 -W 0 -S 240 /dev/hdb

Regards, Dave Hodgins

--
Change nomail.afraid.org to ody.ca to reply by email.
(nomail.afraid.org has been set up specifically for
use in usenet. Feel free to use it yourself.)

On Sun, 08 Apr 2007 15:29:40 +0000, Aardvark wrote:
[color=blue]
> LOL no it's a desktop PC. Since I first started using Mdv and was
> editing
> xorg.conf I wondered why the hell I had tablet PC entries. Just assumed
> they were standard. They bring up errors but don't (seem to) affect
> performance. Would it be ok just to delete these entries altogether?[/color]

Don't take too drastic a step. Just comment them out. I do it like this
to distinguish normal comments from what I change...

Lets suppose I just changed this section...

# Any number of ServerLayout sections may be present. Each describes
# the way multiple screens are organised. A specific ServerLayout
# section may be specified from the X server command line with the
# "-layout" option. In the absence of this, the first section is used.
# When now ServerLayout section is present, the first Screen section
# is used alone.

## Section "DRI"
## Mode 0666
## EndSection

The double asterisks would signify what I just changed.

--
Linux Help: [url]http://rsgibson.com/linux.htm[/url]
Email - [email]rsgibson@verizon.borg[/email]
Replace borg with net

On Mon, 09 Apr 2007 12:37:00 -0400, David W. Hodgins wrote:

[color=blue]
>
> Pay attention on the first boot after installing a new kernel, ensuring
> the video drivers are installed/selected properly.
>[/color]

I had done this when I updated the kernel from 2.6.17-5 to 2.6.17-11.
[color=blue]
> Also, if you had a hard restart (whatever the cause<G>), I would boot
> from a cd, or other running linux system, and fsck all of the filesystems
> related to the failing install.
>[/color]

Done from Ubuntu (the other OS on this machine). Unmounted, fsck'd and
remounted. All clean. Does fsck always happen instantaneously? I thought
it would take some time like MS Scandisk or Chkdsk.
[color=blue]
> If you're using an ext filesystem, check the lost+found, to see if there
> are any obvious files that have been lost.
>[/color]

lost+found in Mandriva is empty. Always wondered what that directory was
and what the name meant :-)
[color=blue]
> I highly recommend turning off hard drive write caching, as the small
> performance gain, is greatly outweighed by the potential loss caused
> by a hard reset.[/color]

If write caching is set on my machine it certainly wasn't me who set it.
How would I know?
[color=blue]
>
> I have the following in /etc/rc.d/rc.local ...
> hdparm -a 8 -d 1 -r 0 -k 1 -u 1 -m 16 -c 1 -A 1 -W 0 -S 240 /dev/hda
> hdparm -a 8 -d 1 -r 0 -k 1 -u 1 -m 16 -c 1 -A 1 -W 0 -S 240 /dev/hdb
>[/color]

Too far beyond my present knowledge so I think for the moment I'll leave
that well alone :-)
[color=blue]
> Regards, Dave Hodgins[/color]

Cheers, Steve Doherty

On Tue, 10 Apr 2007 10:33:05 -0400, Aardvark <aardvark@youllnever.know> wrote:
[color=blue]
> lost+found in Mandriva is empty. Always wondered what that directory was
> and what the name meant :-)[/color]

That's where fsck puts stuff it believes belongs to a file, but can't
figure out which one.
[color=blue]
> If write caching is set on my machine it certainly wasn't me who set it.
> How would I know?[/color]

sudo hdparm -I /dev/hda
will have an asterisk beside "Write cache" if it's turned on.[color=blue][color=green]
>>
>> I have the following in /etc/rc.d/rc.local ...
>> hdparm -a 8 -d 1 -r 0 -k 1 -u 1 -m 16 -c 1 -A 1 -W 0 -S 240 /dev/hda
>> hdparm -a 8 -d 1 -r 0 -k 1 -u 1 -m 16 -c 1 -A 1 -W 0 -S 240 /dev/hdb[/color]
> Too far beyond my present knowledge so I think for the moment I'll leave
> that well alone :-)[/color]

I have losts of stuff in there that doesn't apply, and I should have
make clear which parameters are specific to turning off write caching.
hdparm -W 0 /dev/hda
is all you need to turn off the write caching. One line per drive.

Regards, Dave Hodgins

--
Change nomail.afraid.org to ody.ca to reply by email.
(nomail.afraid.org has been set up specifically for
use in usenet. Feel free to use it yourself.)

On 2007-04-10, Aardvark <aardvark@youllnever.know> wrote:[color=blue]
> On Mon, 09 Apr 2007 12:37:00 -0400, David W. Hodgins wrote:
>
>[color=green]
>>
>> Pay attention on the first boot after installing a new kernel, ensuring
>> the video drivers are installed/selected properly.
>>[/color]
>
> I had done this when I updated the kernel from 2.6.17-5 to 2.6.17-11.
>[color=green]
>> Also, if you had a hard restart (whatever the cause<G>), I would boot
>> from a cd, or other running linux system, and fsck all of the filesystems
>> related to the failing install.
>>[/color]
>
> Done from Ubuntu (the other OS on this machine). Unmounted, fsck'd and
> remounted. All clean. Does fsck always happen instantaneously? I thought
> it would take some time like MS Scandisk or Chkdsk.[/color]

snip

A fsck is not instantaneous. If you mount the partitions of the failing
system in the one that works, what can you see on them using your file
manager? Anything at all?

--
-- ^^^^^^^^^^
-- Whiskers
-- ~~~~~~~~~~

On Sun, 08 Apr 2007 12:36:53 +0000, Aardvark wrote:
[color=blue]
> Mdv 2007 Free, KDE 3.5.
> Went to bed last night and left the PC happily running although the
> monitor was switched off. The wife woke me this morning asking me what my
> password for the computer was. Thinking this was a strange question (when
> the machine is on it logs automatically into my identity so no p/w
> required).
>
> When I got in front of the computer it was sitting at the login screen, so
> I typed in my username and password and hit the enter button. I got the
> nVidia splash screen and then was taken straight back to the login screen.
>
> I've rebooted, tried using different sessions, even tried to log into my
> son's identity (set up on this machine) and all to no avail. All I get is
> the login screen.
>
> I have no idea where to look or what to do to fix this. Totally stumped.
> Can someone please tell me how to trace whatever might be at fault? I'm at
> present using Ubuntu (this machine is dual boot Mdv/Ubuntu) so it's
> definitely not a hardware problem.
>
> The only change to my Mandriva installation was on Friday I upgraded my
> kernel from 2.6.17-5 to 2.6.17-11, also installed 2.6.17-11 kernel source
> and reinstalled my nVidia drivers against the new kernel source.
>
> Any help gladly accepted.
>
> Aardvark[/color]

I'm writing this post from my Mandriva 2007 installation running with
kernel version 2.6.17-11!!!!!

I re-read all the posts in this thread time after time (hell I could
probably quote some of them verbatim, tech-speek 'n all :-)) and although
it wasn't anything conscious something told me to boot Mandriva to run
level 3 and reinstall the nVidia drivers.

At one stage during the installation I got a message that
'/usr/lib/xorg/modules/extensions/libglx.so is not a symbolic link'
although the installation ran to completion. I then rebooted and was
greeted by my KDE desktop.

It was a gut feeling based on the reading of all the posts in the thread
(my own included) that made me go for a driver reinstall, although it was
Whiskers' original posts that pushed me towards my final decision.

All I can say is a big thank you to all who gave me advice and help. You
are, each and every one of you, marvels :-)

Steve Doherty/Aardvark


--
Registered Linux User 413057.
Both Mandriva 2007 and Ubuntu 6.06
You can have it all. My empire of hurt.

Quoth Aardvark :
[color=blue]
> something told me to boot Mandriva to run
> level 3 and reinstall the nVidia drivers.[/color]

Was that 'something' the nVidia README?

<smirk>
--
The Man in the Yellow Hat
Linux with a monkey, since 1996.

On Tue, 10 Apr 2007 16:10:23 +0000, Big Yellow Hats wrote:
[color=blue]
> Quoth Aardvark :
>[color=green]
>> something told me to boot Mandriva to run
>> level 3 and reinstall the nVidia drivers.[/color]
>
> Was that 'something' the nVidia README?
>
> <smirk>[/color]

Why would it be? I haven't even looked at that since I spent the best part
of two days studying it before I first installed the nVidia drivers
shortly after I first ran Mandriva. I've re-installed the drivers a number
of times since, always from run level 3.

That's not to say that something in the README might not have had a
bearing on my decision to re-install the drivers.

I've always been an intuitive person and it was intuition based on reading
and re-reading the posts in the thread without thinking about what I was
reading and possibly the fact that I read the README file a while ago that
gave me the strong feeling that reinstalling the drivers for the second
time against my new kernel source would work. And it did.

I can never explain this, but I know that if I have a strong feeling that
a particular course of action is the correct one, I trust that feeling and
take that course of action. A lot more often than not, it seems to work.

--
Registered Linux User 413057.
Both Mandriva 2007 and Ubuntu 6.06
You can have it all. My empire of hurt.

Aardvark wrote:[color=blue]
> On Tue, 10 Apr 2007 16:10:23 +0000, Big Yellow Hats wrote:
>[color=green]
>> Quoth Aardvark :
>>[color=darkred]
>>> something told me to boot Mandriva to run
>>> level 3 and reinstall the nVidia drivers.[/color]
>> Was that 'something' the nVidia README?
>>
>> <smirk>[/color]
>
> Why would it be? I haven't even looked at that since I spent the best part
> of two days studying it before I first installed the nVidia drivers
> shortly after I first ran Mandriva. I've re-installed the drivers a number
> of times since, always from run level 3.
>
> That's not to say that something in the README might not have had a
> bearing on my decision to re-install the drivers.
>
> I've always been an intuitive person and it was intuition based on reading
> and re-reading the posts in the thread without thinking about what I was
> reading and possibly the fact that I read the README file a while ago that
> gave me the strong feeling that reinstalling the drivers for the second
> time against my new kernel source would work. And it did.
>
> I can never explain this, but I know that if I have a strong feeling that
> a particular course of action is the correct one, I trust that feeling and
> take that course of action. A lot more often than not, it seems to work.
>[/color]
Joining the thread at this point may be pointless :), but did
you have the exact kernel source to match your kernel, was the
symbolic link to /usr/src/linux pointing to the new source for
the kernel (e.g. for the default location
/usr/src/linux-2-6-whatever), and was the new kernel running
when you installed the new drivers? And, of course, did you run
install as root or using sudo?

If any of these is out of kilter, the install either will abort
or you will get drivers that will not work.

Cheers!

jim b.

--
Unix is not user-unfriendly; it merely
expects users to be computer-friendly.

On Tue, 10 Apr 2007 21:58:25 +0000, James D. Beard wrote:
[color=blue]
> Joining the thread at this point may be pointless :), but did
> you have the exact kernel source to match your kernel,[/color]

Yup.
[color=blue]
> was the
> symbolic link to /usr/src/linux pointing to the new source for
> the kernel (e.g. for the default location
> /usr/src/linux-2-6-whatever),[/color]

Yup. I had edited GRUB.
[color=blue]
>and was the new kernel running
> when you installed the new drivers?[/color]

Yup.
[color=blue]
>And, of course, did you run
> install as root or using sudo?
>[/color]

Root.
[color=blue]
> If any of these is out of kilter, the install either will abort
> or you will get drivers that will not work.
>[/color]

I know :-)
[color=blue]
> Cheers!
>
> jim b.
>[/color]

Cheers Steve D.



--
Registered Linux User 413057.
Both Mandriva 2007 and Ubuntu 6.06
You can have it all. My empire of hurt.