General discussion


How to recognize a HD under SCO Unix

By luc.strobbe ·
We 're experiencing a problem with a Unix server (a very old Compaq Proliant). In a RAID-system we found 5 HD. We thought that all the drives were involved in RAID5; so we could swap one of them if it got broke. Unfortunately this is not the case. We found out that there's one boot-HD; three HD in RAID5 and one we couldn't figger out yet. Naturally it is the boot-HD (including the Unix-OS) that died. We ordered a new HD and installed the original version of Unix (Network Sys Release 3.0). Now we're searching for a way to connect the three HD in RAID5 to the new HD with the new installation of the old OS. This way we hope to find our data without having to restore them from backup.

Luc Strobbe,
city-hall Sint-Niklaas

This conversation is currently closed to new comments.

Thread display: Collapse - | Expand +

All Comments

Collapse -

Recognizing HDs in SCO UNIX

by lkarnis In reply to How to recognize a HD und ...


There is a good chance you can get access to all of the data on your RAID 5 array. The command that will let you do the job is:

# mkdev hd

This is a script that configures your kernel to use new hard disk drives. It is a bit of an odd-ball command in that it must accomplish two tasks:

1. Update the Kernel's scsi device table to indicate the address of the new 'hard disk'. (File is /etc/conf/cf.d/mscsi)

2. Take you through a dialogue to optionally partition, format and mount your partitions.

Before you get started, you need to know the exact address (SCSI-wise) of your RAID-5 array. Unfortunately, I can only provide general guidance here. You need to read the driver documentation that came with your Compaq Smart-Array controller. Often, RAID arrays are addressed as LUNs. So it is possible that your first RAID array is SCSI Target 0, LUN 1 (or something like that).

ONce you find the address, you would need to run

# mkdev hd

And insert all of the correct information (controller, bus, scsi target, LUN, etc.). Save and exit. You will then be asked if you want to rebuild the kernel - say yes.


Once the machine comes up, run the command for a 2nd time:

# mkdev hd

Enter the same SCSI info you entered above. If all goes well, you will be asked to:

- set the number of bad blocks to reserve
- determine how much of the disk is to be used for UNIX (default = all)
- do a surface scan (skip this)
- partition the disk

It should show you your existing partitions. If everything is OK, you should see the partition sizes, start/end cyl, file system type, etc. but NOT the partition name. You would need to enter that. Also, it is really important that you ensure that your new partitions ARE NOT to be formatted. When you assign a name, SCO creates entries in /dev of that name (e.g.: if your name was app, the tool would make /dev/app and /dev/rapp).

You would then cross your fingers and select install.

If all goes well, you should be able to use your new file systems. Make the mount-point directories and try mounting them by hand:

# mkdir /app
# mount /dev/app /app

Again, if all goes well, you should be able to see your old data. Next, check out /etc/default/filesys to ensure there is an entry to automatically mount your file system on boot.

If you get all of this right - you should be back in business. If not, you may need to start from scratch and restore your data from back up.

Sorry I can't give you more detail but I'm travelling this week and I don't have access to a SCO box.

Good luck.

Larry Karnis

Related Discussions

Related Forums