Complete Kernel Recovery Walkthrough for SCO OpenServer Administrators

SCO OpenServer Kernel Recovery: Tools, Commands, and Best Practices

Recovering a damaged or non-booting kernel on SCO OpenServer requires clear diagnostics, the right tools, and careful execution. This guide walks through recovery tools, essential commands, and best practices to restore service with minimal downtime.

Overview

SCO OpenServer boots dependably on a healthy kernel and matching kernel modules. Kernel issues typically present as boot hangs, unexpected panics, missing devices, or filesystem errors. The recovery process below assumes local console access (or serial console) and availability of OpenServer installation or rescue media.

Preparation checklist

• Backup: Ensure recent backups of /dev, /etc, /stand, and important configs (if possible before repair).
• Rescue media: Have the original SCO OpenServer installation or rescue floppy/CD/ISO ready.
• Separate storage: Mount or attach external media for saving logs and critical files.
• Boot access: Confirm you can access single-user mode or a rescue shell via installation media or bootloader.

Recovery tools

• Installation/rescue media (floppy, CD, ISO)
• dd, tar, cpio — for copying and restoring filesystems and device files
• mknod — to recreate device nodes in /dev if missing
• fsck (fsck -y or specific fsck.*) — repair filesystems before kernel changes
• mkdev, switchover tools (if present in your OpenServer flavor)
• slogin/ssh via alternate kernel or network-boot environment (if available)
• Hex editor / strings — to inspect kernel image if needed
• Tape/USB backup utilities — for restoring archived kernel or /stand contents

Common recovery scenarios & commands

1) Repair filesystem before kernel restore

If filesystems are damaged, restore them first:

1. Boot from rescue media and get a shell.
2. Identify partitions:

Code
Copy CodeCopied
fdisk -v 

3. Run filesystem checks:

Code
Copy CodeCopied
fsck -y /dev/rdsk/

Repeat for all affected slices.

2) Kernel missing or corrupted in /stand

If the kernel image (commonly in /stand) is missing or corrupted:

1. Mount the root partition read-write (if mounted read-only):

Code
Copy CodeCopied
mount -o remount,rw / 

2. Replace kernel from backup or installation media. Example copying from CD mounted on /mnt:

Code
Copy CodeCopied
cp /mnt/stand/kernel /stand/ cp -pr /mnt/stand//stand/ 

3. Ensure permissions:

Code
Copy CodeCopied
chmod 755 /stand/kernel 

4. Reboot:

Code
Copy CodeCopied
init 6 

3) Recreate /dev device nodes

Missing device nodes can prevent boot or mounting. From rescue shell:

1. Recreate essential nodes (example):

Code
Copy CodeCopied
mknod /dev/console c 0 0 mknod /dev/tty1 c 4 1 mknod /dev/hd0a b 3 0 

2. Restore /dev from backups if available:

Code
Copy CodeCopied
cd /dev tar xvf /path/to/dev-backup.tar 

4) Boot into single-user mode or alternate kernel

• At the bootloader prompt, select single-user or “Safe Mode” to run repair commands without full multiuser services.
• Use an alternate kernel from /stand if multiple kernels exist:

Code
Copy CodeCopied
boot -c “root=/dev/hd0a -s” 

(Replace with your bootloader/kernels syntax.)

5) Restoring from tape or archive

If you have an archived /stand or kernel image:

Code
Copy CodeCopied
cd /stand tar xvf /dev/rmt/0 

Or use dd for raw images:

Code
Copy CodeCopied
dd if=/dev/rmt/0 of=/stand/kernel 

Troubleshooting kernel panics & logs

• Inspect /var/adm/messages and system console output saved on rescue media.
• Use strings on kernel image to check version or corruption:

Code
Copy CodeCopied
strings /stand/kernel | grep -i “OpenServer” 

• If panics reference specific drivers, boot with those drivers disabled or use a minimal kernel.

Best practices

• Always keep verified backups of /stand, /dev, and /etc. Store copies off-host.
• Test restores periodically in a staging environment.
• Keep rescue media current and note exact OpenServer versions and patches applied. Kernel images must match installed modules.
• Document hardware mappings and device major/minor numbers for device recreation.
• Avoid in-place edits on a live /stand without backups; copy then modify.
• Use serial console logging to capture early boot messages for postmortem.
• Label and version kernels in /stand (e.g., kernel-5.0.0) to avoid confusion.
• Plan maintenance windows for kernel replacements and reboots; notify stakeholders.

Quick recovery checklist (compact)

1. Boot rescue media → get shell.
2. Run fsck on all slices.
3. Restore /stand/kernel from backup or installation media.
4. Recreate critical /dev nodes.
5. Boot single-user mode, test, then full reboot.
6. Review logs, verify services and backups.

When to escalate

• Hardware faults (disk/controller failures).
• Kernel image cannot be validated or drivers cause persistent panics.
• No valid backups of /stand or /dev.

If needed, gather: system console logs, /var/adm/messages, kernel image checksum, hardware model and disk identifiers before contacting SCO/OpenServer support or a specialist.

This guide provides pragmatic steps to recover an SCO OpenServer kernel. Follow backups and test procedures to minimize downtime.