|
Home > Archive > Red Hat Installation > May 2006 > Please Help! - system crashes (only) when booting to updated kernel
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
Please Help! - system crashes (only) when booting to updated kernel
|
|
|
| Hi,
I have never had to go to the extreme of posting on Usenet to solve a
problem, but I have been struggling with this for weeks, and no other
posts have helped me. In any case, here is the relevant system
information:
System: Dell PowerEdge 600SC
RAID Card: LSI Logic CERC ATA/4ch RAID
(Series 511 REV C2, BIOS Ver. 1.03)
RAID Level: 5
Logical Drives: 1
Correct Driver: megaraid
Detected Driver: megaraid_mbox
Filesystem setup: (IDE RAID card shows logical drives as SCSI)
/dev/sda1: /boot
/dev/sda2: /data
/dev/sda3: /
/dev/sda4: SWAP
I am trying to get any Fedora above FC2 working on it, and have tried
FC3, FC4, and FC5. The problem is that both FC3 and FC4 detect the
wrong card during installation, and that driver doesn't allow the setup
program to access the logical drive. FC5 doesn't even offer the driver
when the "noprobe" option is used, so I haven't made that work at all.
I can get FC3/FC4 installed as long as I use the "noprobe" option and
manually enter the correct driver. For some reason, both FC3 and FC4
identify the RAID card as (megaraid_mbox), which causes the
installation to fail when it comes time to set up partitions. As long
as I use "linux noprobe" and specify the (megaraid) driver instead of
(megaraid_mbox) at the beginning of the installation, everything goes
like clockwork. That is, until I try to update the thing.
I have been using YUM to update linux lately, but I have also tried
up2date on this system, with the same results. I suspect RPM would
have the same problem. As soon as any new kernel package is installed,
I cannot boot the system to the new kernel. If I select the old kernel
from the grub menu, I can boot the system just fine as before, but of
course, that means still using the old kernel.
My guess about what is happening is that something in the installation
program for the new kernel is causing the wrong driver, probably
(megaraid_mbox) instead of (megaraid) to be set up as the hard disk
driver for boot. If I set the system up on a normal IDE drive,
connected to the system board instead of the RAID card, the updates run
flawlessly, so I know it's a problem between this RAID card and Fedora.
Here is the error message I get after installing a new kernel and
trying to boot to that kernel:
Uncompressing Linux... OK, booting the kernel.
Red Hat nash version 4.2.15 starting
mkrootdev: label / not found
mount: error 2 mounting ext3
ERROR opening /dev/console!!!!:2
error dup2'ing fd of 0 to 0
error dup2'ing fd of 0 to 1
error dup2'ing fd of 0 to 2
switchroot: mount failed: 22
Kernel panic - not syncing: Attempted to kill init!
Some of the things I have already tried before installing new kernel
are:
Resetting each partition name using mklabel
Reformatting SWAP filesystem with version 2 type swap
Remaking initrd images (though not positive I did it right)
Changing reference to "megaraid_mbox" to "megaraid" in
modprobe.conf & hwconf
Renaming megaraid_mbox.ko and copying megaraid.ko to that filename
Installing everything (except SWAP) on one partition
Copying old kernel config and initrd files to new name of new
kernel files
None of those worked. So, needless to say, I need a lot of help. I'm
all out of ideas, so I'm asking the gurus out there. Hopefully someone
can help me. I think the most important point is that the new kernel
installation is what is bad, somehow not incorporating the correct
driver, since the original installation remains good, even after the
new kernel is installed. I'm hoping it's something simple, but *any*
help would be greatly appreciated. I will answer any further questions
about my configuration if asked. This thing is driving me crazy!
Thanks!!!
| |
|
| Nicholas Andrade wrote:
> Have you considered just recompiling the kernel yourself? You can use
> the kernel source patched by Red Hat (see the Fedora Release Notes on
> how to extract the kernel source from the SRPM) or you can use a vanilla
> one from kernel.org. If you know the hardware well, this should be the
> best option. If you must use the precompiled kernels in the yum
> repository, then could you post the contents of your /boot/grub/grub.conf.
Recompiling the linux kernel is something I need to learn how to do at
some point, since I've never done it and have been playing with the OS
for years now. I'd rather do it under better circumstances, though.
Here is the info you asked for (the grub.conf file). I included even
the standard commented-out stuff at the beginning of the file, and I've
listed it for two configurations. The first one is the grub.conf that
I had originally, with the 4-partition setup I talked about in my
original post. The second one is the one I currently have, since the
last installation was the one where I put everything except SWAP on one
partition (no /boot). They're both the same in that the second option
(the original kernel) works if selected, but the new kernel (selected
by default) does not.
------
First grub.conf - from this partition setup:
sda1 /boot
sda2 /data
sda3 /
sda4 (swap)
------
# grub.conf generated by anaconda
#
# Note that you do not have to rerun grub after making changes to this
file
# NOTICE You have a /boot partition. This means that
# all kernel and initrd paths are relative to /boot/, e.g.
# root (hd0,0)
# kernel /vmlinux-version ro root=/dev/sda3
# initrd /initrd-version.img
# boot=/dev/sda
default=0
timeout=5
splashimage=(hd0,0)/grub/splash.xpm.gz
hiddenmenu
title Fedora Core (2.6.16-1.2096_FC4)
root (hd0,0)
kernel /vmlinuz-2.6.16-1.2096_FC4 ro root=LABEL=/ rhgb quiet
initrd /vmlinuz-2.6.16-1.2096_FC4.img
title Fedora Core (2.6.11-1.1369_FC4)
root (hd0,0)
kernel /vmlinuz-2.6.11-1.1369_FC4 ro root=LABEL=/ rhgb quiet
initrd /vmlinuz-2.6.11-1.1369_FC4.img
(NOTE: Choosing option #2 boots, choosing option #1 gives the error
from the initial post)
---------------------------------------
This config is the one where I put most of the system on one partition.
The last install I did had a /boot partition, so the paths were
different. Also, that version of the new kernel was 2.6.16-1.2096_FC4
instead of the newer 2.6.16-1.2108 from this config.
------
Second grub.conf - from this partition setup:
sda1 (blank)
sda2 /mnt/data
sda3 /
sda4 (swap)
------
# grub.conf generated by anaconda
#
# Note that you do not have to rerun grub after making changes to this
file
# NOTICE You do not have a /boot partition. This means that
# all kernel and initrd paths are relative to /, e.g.
# root (hd0,2)
# kernel /boot/vmlinux-version ro root=/dev/sda3
# initrd /boot/initrd-version.img
# boot=/dev/sda
default=0
timeout=5
splashimage=(hd0,2)/boot/grub/splash.xpm.gz
hiddenmenu
title Fedora Core (2.6.16-1.2108_FC4)
root (hd0,2)
kernel /boot/vmlinuz-2.6.16-1.2108_FC4 ro root=LABEL=/ rhgb quiet
initrd /boot/vmlinuz-2.6.16-1.2108_FC4.img
title Fedora Core (2.6.11-1.1369_FC4)
root (hd0,2)
kernel /boot/vmlinuz-2.6.11-1.1369_FC4 ro root=LABEL=/ rhgb quiet
initrd /boot/vmlinuz-2.6.11-1.1369_FC4.img
(NOTE: The exact same problem as with the other configuration -
choosing option #2 boots, choosing option #1 gives the error from the
initial post)
------
It doesn't seem to matter whether the /boot partition exists nor what
kernel version is installed. The initial installation program must
have done to the system configuration that none of the update programs
do. With this in mind, I don't even know if a compiled kernel would do
anything. Does the installation CD compile the kernel or does it stick
a standard kernel in there. I'm sure the problem can be found by
looking at the differences between a CD install (using "linux noprobe")
and a yum/up2date/rpm kernel update.
Thanks again!!!
| |
| Michael Heiming 2006-05-25, 7:18 am |
| In comp.os.linux.misc ZZYZX <zzyzx2000@gmail.com>:
[ problems with Fedora Core on Dell PowerEdge ]
> System: Dell PowerEdge 600SC
This looks like a server. I'd suggest to try out RHEL 4 (update 3
is current (iirc)) or one of the clones like CentOS if you don't
need/want the support from RH and see if this works better. Be
sure to download the latest update CD/DVD and run 'yum update'
after installing or 'up2date -u' if you go for RHEL.
[stuff]
--
Michael Heiming (X-PGP-Sig > GPG-Key ID: EDD27B94)
mail: echo zvpunry@urvzvat.qr | PERL -pe 'y/a-z/n-za-m/'
#bofh excuse 92: Stale file handle (next time use
Tupperware(tm)!)
| |
|
| Michael Heiming wrote:
> In comp.os.linux.misc ZZYZX <zzyzx2000@gmail.com>:
>
> [ problems with Fedora Core on Dell PowerEdge ]
>
>
> This looks like a server. I'd suggest to try out RHEL 4 (update 3
> is current (iirc)) or one of the clones like CentOS if you don't
> need/want the support from RH and see if this works better. Be
> sure to download the latest update CD/DVD and run 'yum update'
> after installing or 'up2date -u' if you go for RHEL.
>
> [stuff]
>
> --
> Michael Heiming (X-PGP-Sig > GPG-Key ID: EDD27B94)
> mail: echo zvpunry@urvzvat.qr | PERL -pe 'y/a-z/n-za-m/'
> #bofh excuse 92: Stale file handle (next time use
> Tupperware(tm)!)
Thanks. Luckily, I do have a copy of RHEL V4. It's not licensed to me
so I can't use it forever, but it will be at least a month before my
friend (the license owner) will install it on his server. I'll get his
permission and attempt to install it and let you all know. I'll
install it using the 4-partition setup that I originally had on my
Fedora system, that is to say:
sda1 = /boot
sda2 = /data
sda3 = /
sda4 = (swap)
I'll get that going over the next few days, provided someone doesn't
suggest a better/easier solution by then. Even if I can get RHES to
work, I'll still have to figure out how it did it (and how to make
Fedora do it), as I will only be borrowing the copy of RHES. I don't
quite have the cash to justify spending $700 (for my own copy of RHEL)
even if it does work. I have a copy of Win2000 Server that belongs to
me, but I'd much rather do it with Linux!
Thanks again!
ZZ
| |
|
| OK, I'll get going with that plan over the next few days. If I have
any problems I suppose I can ask you folks
I did try installing RH Enterprise Server, and - get this - it detected
my RAID card without my having to use the "linux noprobe" option. That
is to say, it detected my card *correctly* with no more problems. I'm
fairly confident that it would have applied updates correctly as well.
I guess that's the difference $700 makes. Since I don't have that, and
I refuse to use Windows when I have the option, I'll start compiling.
Thanks for all the help. I'll post again when I get it working
ZZYZX
| |
| Michael Heiming 2006-05-29, 5:30 pm |
| In comp.os.linux.misc ZZYZX <zzyzx2000@gmail.com>:
> OK, I'll get going with that plan over the next few days. If I have
> any problems I suppose I can ask you folks
> I did try installing RH Enterprise Server, and - get this - it detected
> my RAID card without my having to use the "linux noprobe" option. That
> is to say, it detected my card *correctly* with no more problems. I'm
> fairly confident that it would have applied updates correctly as well.
> I guess that's the difference $700 makes. Since I don't have that, and
> I refuse to use Windows when I have the option, I'll start compiling.
You just need to use CentOS as already pointed out to you, it is
a *free* clone of RHEL. Sorry but I sometimes have the feeling
people don't really read what others write? Or is it to difficult
to type "CentOS" into google, it's really great if used for what
it was made for.
It is in addition beyond me why you use groups.google which is
the worst interface to write on usenet, when your ISP seems to
have his own nntp server. At least news.charter.com resolves
quite fine.
Anyway, perhaps I'm just getting old? ;-)
Good luck
BTW
Please remember to quote context, this is usenet *not* some
groups.google forum.
--
Michael Heiming (X-PGP-Sig > GPG-Key ID: EDD27B94)
mail: echo zvpunry@urvzvat.qr | PERL -pe 'y/a-z/n-za-m/'
#bofh excuse 113: Root nameservers are out of sync
| |
|
| Michael Heiming wrote:
> In comp.os.linux.misc ZZYZX <zzyzx2000@gmail.com>:
>
> You just need to use CentOS as already pointed out to you, it is
> a *free* clone of RHEL. Sorry but I sometimes have the feeling
> people don't really read what others write? Or is it to difficult
> to type "CentOS" into google, it's really great if used for what
> it was made for.
>
> It is in addition beyond me why you use groups.google which is
> the worst interface to write on usenet, when your ISP seems to
> have his own nntp server. At least news.charter.com resolves
> quite fine.
>
> Anyway, perhaps I'm just getting old? ;-)
>
> Good luck
Sorry if it offended you, but I used Google instead of Charter's nntp
server for practical reasons. This the first Usenet post I've made in
years, and possibly my last for years. Since I already had a Gmail
accout, it was quicker than calling Charter to change my ISP password,
which someone else at my location set up before leaving the company.
For my purposes, a shitty interface is preferable to waiting 45 minutes
for a tech. Maybe you would wait for the tech, I dunno.
If I decide to post here a lot, I'll go back to NNTP.
ZZ
| |
| Michael Heiming 2006-05-29, 5:30 pm |
| In comp.os.linux.misc ZZYZX <zzyzx2000@gmail.com>:
> Michael Heiming wrote:
[stuff]
[vbcol=seagreen]
> Sorry if it offended you, but I used Google instead of Charter's nntp
> server for practical reasons. This the first Usenet post I've made in
> years, and possibly my last for years. Since I already had a Gmail
> accout, it was quicker than calling Charter to change my ISP password,
> which someone else at my location set up before leaving the company.
> For my purposes, a shitty interface is preferable to waiting 45 minutes
> for a tech. Maybe you would wait for the tech, I dunno.
;-)
> If I decide to post here a lot, I'll go back to NNTP.
Strong point!
Not a real issue, to get back to your problem, you can turn RHEL
into CentOS without reboot, there should be instructions on
howto do this task on centos.org. Just be sure to run 'yum
update' after doing so.
Good luck
--
Michael Heiming (X-PGP-Sig > GPG-Key ID: EDD27B94)
mail: echo zvpunry@urvzvat.qr | PERL -pe 'y/a-z/n-za-m/'
#bofh excuse 346: Your/our computer(s) had suffered a memory
leak, and we are waiting for them to be topped up.
|
|
|
|
|