Boot hangs on modprobe errors

Discussion:

Laura Speck

2008-10-02 18:34:36 UTC

Hi all,

I have a fresh install of Fedora 9 on a Dell poweredge 2650. Install
went fine, no errors or warnings. Rebooted it, and it hangs after:

serial 00:06: unable to assign resources
Red Hat nash version 6.0.52 starting
INFO: task modprobe:<number here that I cannot remember> blocked for
more than 120 seconds
"echo 0> /proc/sys/kernel/hung_task_timeout_secs" disables this message.
INFO: task scsi_scan_0:449 blocked for more than 120 seconds
"echo 0> /proc/sys/kernel/hung_task_timeout_secs" disables this message.

We used 32bit fedora 9 to install onto a 3 dish scsi raid5. I am not
sure if this machine is 32bit or 64bit, but I figured 32bit fedora
should work on both?

Rebooted again, and now it's not doing anything. It gave the error
about the serial, then started nash up, but now nothing. And it's been
a half hour.

I am completely clueless right about now and any help would be greatly
appreciated, even a point in the right direction :)

--
Laura Speck

Craig White

2008-10-02 18:51:52 UTC

Permalink

Post by Laura Speck
Hi all,
I have a fresh install of Fedora 9 on a Dell poweredge 2650. Install
serial 00:06: unable to assign resources
Red Hat nash version 6.0.52 starting
INFO: task modprobe:<number here that I cannot remember> blocked for
more than 120 seconds
"echo 0> /proc/sys/kernel/hung_task_timeout_secs" disables this message.
INFO: task scsi_scan_0:449 blocked for more than 120 seconds
"echo 0> /proc/sys/kernel/hung_task_timeout_secs" disables this message.
We used 32bit fedora 9 to install onto a 3 dish scsi raid5. I am not
sure if this machine is 32bit or 64bit, but I figured 32bit fedora
should work on both?
Rebooted again, and now it's not doing anything. It gave the error
about the serial, then started nash up, but now nothing. And it's been
a half hour.
I am completely clueless right about now and any help would be greatly
appreciated, even a point in the right direction :)

----
I have CentOS 5 (previously 4) running on a PE 2650 which came with a
PERC 3/Di RAID adaptor (Adaptec actually) and it clearly is a 32 bit
system so I would think that you installed the right version.

Obviously the installer ran and that tells you that the 32 bit OS works.

I would suspect that have a grub issue with the install (i.e. - not
installing grub on /dev/sda)

Suggest that you do...

boot installation disk again
enter 'linux rescue' at boot
have it locate the installation for you and get to a command prompt
type 'chroot /mnt/sysimage'
type 'grub-install /dev/sda'
type 'exit'
type 'exit'
remove the installation CD and reboot

Craig

Craig White

2008-10-02 18:58:08 UTC

Permalink

Post by Craig White

----
I have CentOS 5 (previously 4) running on a PE 2650 which came with a
PERC 3/Di RAID adaptor (Adaptec actually) and it clearly is a 32 bit
system so I would think that you installed the right version.
Obviously the installer ran and that tells you that the 32 bit OS works.
I would suspect that have a grub issue with the install (i.e. - not
installing grub on /dev/sda)
Suggest that you do...
boot installation disk again
enter 'linux rescue' at boot
have it locate the installation for you and get to a command prompt
type 'chroot /mnt/sysimage'
type 'grub-install /dev/sda'
type 'exit'
type 'exit'
remove the installation CD and reboot

----
come to think of it...did you add hard drives to the system after the
installation that is confusing things?

Craig

Craig White

2008-10-02 18:58:08 UTC

Permalink

Post by Craig White

----
I have CentOS 5 (previously 4) running on a PE 2650 which came with a
PERC 3/Di RAID adaptor (Adaptec actually) and it clearly is a 32 bit
system so I would think that you installed the right version.
Obviously the installer ran and that tells you that the 32 bit OS works.
I would suspect that have a grub issue with the install (i.e. - not
installing grub on /dev/sda)
Suggest that you do...
boot installation disk again
enter 'linux rescue' at boot
have it locate the installation for you and get to a command prompt
type 'chroot /mnt/sysimage'
type 'grub-install /dev/sda'
type 'exit'
type 'exit'
remove the installation CD and reboot

----
come to think of it...did you add hard drives to the system after the
installation that is confusing things?

Craig

Alan Cox

2008-10-02 20:39:56 UTC

Permalink

Post by Laura Speck
INFO: task scsi_scan_0:449 blocked for more than 120 seconds

It got stuck trying to scan a scsi device - perhaps the RAID array. If
you hit a key during the boot prompt from grub and edit the boot line to
remove the "quiet" it will spew rather more info which should help see
what broke and where

Post by Laura Speck
"echo 0> /proc/sys/kernel/hung_task_timeout_secs" disables this message.
We used 32bit fedora 9 to install onto a 3 dish scsi raid5. I am not
sure if this machine is 32bit or 64bit, but I figured 32bit fedora
should work on both?

Yes, and 64bit will print an error on 32bit.

Alan

Laura Speck

2008-10-02 21:57:18 UTC

Permalink

An HTML attachment was scrubbed...
URL: http://lists.fedoraproject.org/pipermail/users/attachments/20081002/7edc9c99/attachment.html

Craig White

2008-10-02 22:11:17 UTC

Permalink

Post by Alan Cox

Post by Laura Speck
INFO: task scsi_scan_0:449 blocked for more than 120 seconds

aac_srb: aac_fib_send failed with status: 8195
I found this https://bugzilla.redhat.com/show_bug.cgi?id=450444
Adding aacraid.dacmode=0 did not help, same errors. But mem=4G did. I
went into the GRUB menu and edited the boot line ('e') and it booted
fine.
Can anyone point me in the direction of how I would permanently fix
this? I noticed on
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=94cf6ba11b that it says "limiting the discovered kernel memory (mem=4G) with an associated loss in runtime performance" - is there a better way to fix this issue, now that I can actually access the box?
Thanks in advance!!! :) Much appreciated.

----
You probably realize this but you could add the mem=4G line
to /boot/grub/grub.conf to permanently use that option.

You could also try using the PAE kernel as that is capable of addressing

3 GB of RAM

You probably want to make sure the firmware on the PERC 3/Di is the
latest version as suggested by Jeff Lawson since he reported that the
problem went away after updating (and using the PAE kernel but he was
using CentOS 5, not Fedora)

Craig

Craig White

2008-10-02 22:11:17 UTC

Permalink

Post by Alan Cox

Post by Laura Speck
INFO: task scsi_scan_0:449 blocked for more than 120 seconds

----
You probably realize this but you could add the mem=4G line
to /boot/grub/grub.conf to permanently use that option.

You could also try using the PAE kernel as that is capable of addressing

3 GB of RAM

Laura Speck

2008-10-02 21:57:18 UTC

Permalink

An HTML attachment was scrubbed...
URL: http://lists.fedoraproject.org/pipermail/users/attachments/20081002/7edc9c99/attachment-0001.html

Laura Speck

2008-10-02 18:34:36 UTC

Permalink