Pages

Search

Dual-boot machine dies after SUSE install - Forums Linux

Dual-boot machine dies after SUSE install - Forums Linux

Dual-boot machine dies after SUSE install

Posted: 31 Dec 2005 11:37 AM PST

In comp.os.linux.setup Peter T. Breuer <it.uc3m.es>: 

[ system comes to a grinding halt unexpected ]
 
 
 
 

Indeed! CPU fan sounds like the closest from the OP, well
spotted!

BTW
Guten Rutsch!

--
Michael Heiming (X-PGP-Sig > GPG-Key ID: EDD27B94)
mail: echo qr | perl -pe 'y/a-z/n-za-m/'
#bofh excuse 229: wrong polarity of neutron flow

Windows XP upgrade

Posted: 30 Dec 2005 04:04 PM PST

On Sat, 31 Dec 2005 00:04:54 +0000, Ed Doyle wrote:
 

It's pretty much certain that a new Windows install will over-write the
MBR. So you'd then have to boot off something else ( rescue cd, boot
floppy, etc. ) and reinstall/initialize grub. Personally, I like tomsrtbt
- http://www.toms.net/rb/
knoppix also makes a very good rescue tool. I'm not all that familiar
with fedora, but older redhat cd's used to be able to do rescue functions.
 

I have done it a while ago, although I was using lilo as the boot
loader, not grub. You'll also find tons of hits if you google things
like "grub reinstall windows" or similar.
 

If your current Windows installs and linux are on the same physical
disk, and you change the partitioning for windows, this will change the
way the linux partions are numbered...So you'd have to again use a boot
cd/floppy and change your fstab to match the new disk layout. If Linux is
on a different physical disk, or if you left the partioning as-is for
Windows and simply formatted + reinstalled, this wouldn't be an issue.

Make sure you can boot from cd/floppy *before* you mess with
anything.
Make a note of the current disk layout for Linux...Where / is, where your
swap is, etc. Like hda5, hda6, whatever.
Backup whatever is important.
Make sure when (re)installing Windows you are *sure* you
are using/formatting only the windows partitions.
After the WinXP install, boot from cd/floppy, and check what the Linux
partions are seen as now. If they're different, you'd have to edit your
fstab to match, edit your grub config, then reinstall grub. After that,
you should be fine.
IMHO, I think it's better to use seperate physical disks for different
OS's, whenever possible.

--
- Matt -

Simple one: How to set a static IP address?

Posted: 30 Dec 2005 07:51 AM PST

Thommy M. wrote: 
You could do it the simple way and go to
system-->Administration-->Networking
and change it from DHCP to static.

Herbert.

CD Drive not being recognized

Posted: 30 Dec 2005 07:42 AM PST


"Ken K" <headfog.com> wrote in message
news:Ruxtf.1810$.. 

It depends on the OS. Recent RedHat releases detect it with Kudzu and
possibly other tools, and put in entries in the /etc/fstab corresponding to
/media/cdrom/ or /media/cdrecorder/, which refer to appropriately generated
/dev/cdrom or /dev/cdrom1 devices. They've gotten very clever about it!

[help]linux and my webcam

Posted: 30 Dec 2005 01:35 AM PST


"Michael Heiming" <michael+heiming.de> wrote in message
news:heiming.de... 

And if this drives you insane, in the short term, you can also set up an
init script like those in /etc/init.d to load it or not depending on the run
level. I've done this with ide-scsi drivers because I didn't *WANT* ide-scsi
on all the time, it messed with an old system, but was required by old
versions of cdrecord. So I'd use the init script or not depending on what I
was up to.

Moving harddisk with Linux installation to a different machine.

Posted: 29 Dec 2005 07:40 PM PST

On Fri, 30 Dec 2005 04:40:57 +0100, <com> wrote:
 

People are doing it all the time and generally it works perfectly.

There is at least one way it can fail, though: The new hardware may
not (yet) be suppoted by the kernel on the hard disk.

In most cases you will still be able to boot the computer, but perhaps
not be able to start X, so you would be stuck with console mode until a
sufficiently capable kernel is installed. If you know what kernel modules
and versions you need, you may be able to download and install a new
kernel before moving the disk over.

You are using a 2.4 kernel, and I just don't know how much support for
newer hw has been backported to 2.4 kernels. Be aware that you can have
multiple kernels installed. You could have a recent 2.6 kernel on the
disk before the move. But do google to find out if the hw you want to
buy is supported yet. (Hey, everyone, is there any problem with
Redhat glibc/tls/nptl and 2.6 kernels? I believe the 2.6 kernel runs
older libc stuff just fine.)
 

What is too big? The kernel? Yes, that is true. But if the new
computer has a floppy drive you can prepare a grub floppy, and then
you are almost sure to manage to setup the mbr properly afer the move.

(A grub floppy does not contain the kernel, etc, it just contains
grub's stage1 and stage2 files. Booting from it you get a command
prompt that enables you to query the disks, search for files across
partitions, set up the MBR or a partition boot record, invoke a
config file with boot stanzas, edit the stanzas before using them
(but not save the changes to the file), and/or boot specifying your
own boot commands.)

However, floppies are not in vogue anymore. They tend to be left out
from newer computers. It is also possible to setup a Grub CD, but I
have never done that.

Notice, as will be evident from the points below, that running a Grub
CD is not the same thing as running Grub under a Linux Rescue CD or a
Knoppix Live CD.
 

You are right. Grub has a problem determining the right order that the
Bios will enumerate the disks. With a grub floppy, it's easy, you can
use grub to query the disks, and see what number the disks have. Just
use the "find" command while running Grub natively (i.e.not under a Linux
kernel).

If there is only one disk, it is easy too, the disk will be number zero.
Say (hd0) to grub. The problem arises when you run grub under Linux to
setup the mbr. Then grub has no way of talking to the bios, it has to
access the disks through the Linux kernel. If you specify a particular
disk to boot from in the Bios setup, the Bios takes that disk out of the
regular order and places it first, so Grub will see it as (hd0). There
is no way Grub can find out that while running Linux. Otherwise, the
standard IDE disks always get enumerated hda, hdb, hdc, hdd, but missing
disks are skipped, so if hdb is not present, hdc will get the number
after hda. Other disks could come before or after, depending on details
in the Bios of the computer and the Bios extensions in installed (or
on-board) controllers. Again something hard for Grub to know.

If you know the disk is going to be (hd0) on the new computer, while it
is /dev/hdc under linux on the old computer, I believe you can create a
"device.map" file in /boot/grub, where you specify "(hd0) /dev/hdc", and
then run "grub-install /dev/hdc", still under linux, to have the mbr of
the disk set up for the coming configuration. This will not affect the
mbr of /dev/hda, so you should still be able to boot OK on the old
computer. It's like having two pointers to the same area, and the
/dev/hda mbr is used when booting the old computer. The next step is to
edit /boot/grub/menu.lst or */grub.conf (depending on which distro you
have). This file is shared by the two boot paths, and will be correct
for only one of them at the time. But if it is incorrect, you still get
Grub itself loaded during boot, and can give Grub commands to bypass the
incorrect config file. (You can even have two config files, and 1)
specify a non-standard config filename to grub-install, and 2) you can
also say, eg., "configfile (hd0)/grub/newgrub.conf" at the grub prompt
after it fails to boot using an incorrect config file.)

Yet another idea is to have a double set of boot stanzas in the
grub config file. The only thing that fails then, should be the splash
image file. You can't have two paths for it. Expect an uglier boot
menu. Consider using the "fallback" statement in the config file.
(I have never used it.)

Caveat: Few people spend their days rearranging boot disks all the day,
and neither do I. Everything here is kind-of fragile, subject to
misunderstandings or incorrect memories on my part. I have done some
experimenting with Grub, read some of the code, received helpfull
corrections when I have said something wrong here in the ng, etc,
but still the chances of making mistakes are good. I don't have access
to all the possible hw combinations either. The Grub documentation is
notoriously unclear, although it is improving, and it tries to be verbose.

I would like to hear about your experiences, because I am considering
writing something about it, if I can collect enough experiences and
testimonials (bad ones too, but explained and understood ones better).

-Enrique

what do these names stand for

Posted: 29 Dec 2005 09:53 AM PST

"x" <com> writes:
 

They stand for themselves.
Originally /var was for variable, /usr was for user, /etc was for etcetera,
/proc for procedure, init for initial or initialize, /opt for optional.
But those had little significance.
NOw they are names in their own right. they mean what they are defined to
mean.
/var/ is for temporary or log or spool system files.
/usr is for most stuff
/etc/ for setup data
/proc is a special file system which is a view into the kernel (Ie files
here are actually variables in memory in the kernel)
/initrd does not exist on Mandrake
/opt is for user installed programs often.

grub menu oddity

Posted: 29 Dec 2005 07:06 AM PST


Richard wrote: 

hi;
I am very curious as well - perhaps a few more questions may help
Enrique and others, much more knowledgeable about linux than I, figure
this out for you;

-- How many OSes are in your Multi-Boot array ?
-- How Many HDDs ?
-- Any PCI IDE Controller Cards installed or SCSI Adapter Cards ?
-- Any RAID arrays ?
-- System Mobo make/model ?
-- BIOS type and version (Award, Pheonix, AMI, etc)

If there is more than 1 version/copy of grub somewhere else on your
HDDs, is it possible that that 'other' copy is being used after BIOS
config changes ? More-so, is it possible there's a Symbolic link
somewhere that's pointing to a different kernel, or updated kernel ?

There can be issues with the way grub sees the various default linux
numbering scheme, as I've found from reading previous info posted by
Enrique.
 
of issue with the CMOS NVRAM area (nowadays, the NVRAM/RTC is
integrated into the South Bridge itself) and the storage of the changes
made there-in.

Possible a setting in the BIOS is set to Auto "Load Setup Defaults"
after saving changes ? There's usually options to "Save Custom
Defaults" - and "Load Custom Defaults" -- curious what the options you
have.

Have you ever Flashed (updated) the Motherboard BIOS ROM ?

Another quick thought is the CMOS battery is old and dying, though
other issues/problems as a consequence of this would be much more
prevelant.

setup desired behavior for alt-tab

Posted: 28 Dec 2005 08:20 PM PST

On Thu, 29 Dec 2005 07:18:36 +0100, Peter T. Breuer <it.uc3m.es> wrote:
 

There was another post recently with an equivalent question, but the
window manager was Fluxbox. I did look into the Fluxbox documentation,
but I could not find documented how Fluxbox defines the order of the
windows.

If the order is the stacking order, and the window selected by the
:NextWindow command is raised to the top of the stacking order, then,
if :NextWindow is invoked again, the previous top-of-stack window would
be selected.

But if the order is some fixed order of the windows in a table or linked
list inside the window manager, I could not see any way of achieving Windows
semantics with the command set offered in the documentation.

There was a reference to an external tool, wmctrl, which could be used to
write scripts that did more complex operations, but I could not find any
way of querying the window manager about the current stacking order, so
I would only have been able to implement an emulation of the Windows
semantics if *all* operations that change the stacking order are routed
through the external program.

There was an option to list the windows currently managed by the manager,
but the list was the same independent of the window stacking order.

It seems like the easiest way is to hack the source of the window manager
and add a new function, that can subsequently be bound to Alt-Tab. Or, you
could write a script using wmctrl, that would appear to work as desired
provided that no other window raising or lowering operations where made in
the mean time. That script would need to have some method of saving
the current intended stacking order.

However, the full Windows semantics requires that you can bind actions to
bot keypress and key release. You want to be able to hold down the
alt key while pressing repeatedly on the Tab key, to select any window from
the list, and raise it when the alt key is released.

-Enrique

Sound not working

Posted: 28 Dec 2005 08:21 AM PST

I have a dumb question,(I am not running Fedora so it might be
immediately obvious to you); do you have a file on you system called
alsaconf?

That may be the key to getting it going.

Sharing a Printer

Posted: 28 Dec 2005 05:00 AM PST

It's ok i got it working by modifying the mime.conf and mime.types
file. Thanks for your help.

Problems setting up DNS, gateway und subnetmask!

Posted: 27 Dec 2005 04:10 PM PST

In message <phx.az.us>
example.tld (Moe Trin) wrote:
 

I was assuning, if it wasn't clear, that the 10.0.0.0 network was being
defined by the router. Routers, in my experience usually come set with a
class A or C network defined. (Either 10.0.0.0/8 or 192.168.x.0/24). It is
usual, though obviously not essential, to leave the router set and alter the
rest to match. You can of course alter the router to match your network,
provided you aren't using somebody else's network number, i.e. you need to
be on a private range.

The latter solution is simpler if you already have a network, but it can be
a little tricky to set the router up initially, or after a reset. Using
secondary addresses is the way I solve that one.
 

--
Alan Adams, from Northamptonshire
freeserve.co.uk
http://www.nckc.org.uk/

glibc-2.3 code on glibc-2.2 system?

Posted: 24 Dec 2005 02:08 AM PST

Denes Molnar <purdue.edu> wrote:
 

Sure - LD_LIBRARY_PATH and friends.
 

Everyone does this all the time. I run glibc 2.1.
 

Then fix it some more. Find out what else is needed. Smetimes it's not
obvious. You also want to control secondary linkages .. this is not
handled well by LD_LIBRARY_PATH and the solution is usually to run
with ld.so --library-path ... instead.
 

Keep adding. Locate the crash pint and replace that library.
 

It matters.
 

Doesn't matter - you know better than it.

Peter