Problems booting after installing RHEL ES 2.1

[Douglas D. Denney]

Hello,

I have a DL380 server with both a SmartArray 532 PCI card and the built-in 5i card. The 532 is being used to control the disks. The 5i card is unused.

I installed RedHat ES 2.1 on the server, using CDs. All seemed well. The install finished and the server rebooted. As the server was booting, there was a kernel panic and the following messages were displayed (note, the server never successfully booted the OS):
===============================================
VFS: Mounted root (ext2 filesystem).
Red Hat nash verSCSI subsystem driver Revision: 1.00

sion 3.2.6 startCompaq CISS Driver (v 2.4.27)

ing
Loading scsi_mod module
Loading sd_mod modcciss: using DAC cycles

ule
Loading cciblk: queue c03e6dc0, no I/O memory limit

ss module
cciss: using DAC cycles

blocks= 35553120 block_size= 512

heads= 255, sectors= 32, cylinders= 4357


blocks= 142253280 block_size= 512

heads= 255, sectors= 32, cylinders= 17433


blk: queue c03e6e5c, no I/O memory limit

Partition check:

cciss/c1d0: p1 p2 p3 p4 < p5 p6 p7 p8 >

cciss/c1d1: p1

Loading jbd moduJournalled Block Device driver loaded

le
Loading ext3 module
Mounting /proc filesystem
Creating root device
Mounting root filesystem
mount: error 19 mounting ext3
pivotroot: pivot_root(/sysroot,/sysroot/initrd) failed: 2
Freeing unused kernel memory: 244k freed

Kernel panic: No init found. Try passing init= option to kernel.
===============================================

I haven't yet had the time to boot from the rescue CD to see if there's anything I can do. But, I wonder if anyone had seen this before or had heard of any issues with the 532 card.

I have 2 other server with only a 5i card, and they installed the same version of the OS, booted, and came up just fine.

Any help would be appreciated.
Thanks.

[Stuart Browne]

It would seem that the boot-loader entry is pointing to the incorrect path.

You'll need to boot off your rescue disks, verify the /dev/cciss/c?d?p? device node for the root= entry is correct (using lilo right? /etc/lilo.conf, and dont forget to re-run '/sbin/lilo'!).

If you feel up to giving us more details (if this is giving you strife), get the contents of the /etc/lilo.conf off your disk, and send it here for us to have a quick look over.

[Douglas D. Denney]

Seems that I may have discovered the solution to my problem. I found information on bugzilla (http://bugzilla.redhat.com), bug ID: 89166.

In this bug description, it describes my problem being related to having my OS disks on controller 1 instead of controller 0 (with the cciss SCSI controllers). This causes the first boot after OS install to fail. The workaround mentioned in the article did indeed work for me. I was able to interrupt the boot process, edit the grub settings and instead of saying this:

root=/dev/cciss/c1d0p3

I put this in:

root=6903

It works. I was amazed that this worked at all, until, after getting the OS booted I discovered that the major/minor numbers for device /dev/cciss/c1d0p3 are (105,3), which is decimal for 6903.

Hopefully this helps someone else in a similar situation with more than 1 cciss scsi controller. Now, all I need is to find the patch for RHEL ES 2.1. All I've seen so far are patches for itanium servers...