topic Re: Getting RH8 to recognize new drive on cciss controller in Operating System - Linux

Getting RH8 to recognize new drive on cciss controller

Johan Sunnerstig — Wed, 17 Sep 2003 06:11:50 GMT

We have a DL360 G3 with a 532 controller, and an attached SW enclosure.

The box is running RedHat 8 with the support pack installed, working fine.

Now, how would I go about getting the OS to find the disk?
Currently there's a logical volume with the device name /dev/cciss/c1d1 with one partition mounted.
So, I'd assume the new virtual drive is /dev/cciss/c1d2

Seems like this assumption is wrong though, the OS can't find anything there.

I've tried doing an "echo "rescan" > /proc/driver/cciss/cciss1", didn't work either.

Any solutions to this, short of restarting the box?

Another question while Im at it.
Is there any way to move a disk that's part of a RAID1 to another box, then just using it as a standalone disk to extract data from?

Regards
Johan

Re: Getting RH8 to recognize new drive on cciss controller

Alexander Chuzhoy — Wed, 17 Sep 2003 06:18:53 GMT

if this logical disk /dev/cciss/c1d1 doesn't have a file system-then create one first
mkfs /dev/cciss/c1d1

then create directory under /mnt for example.
mkdir /mnt/data
Then mount the /dev/cciss/c1d1 under that directory
mount /dev/cciss/c1d1 /mnt/data

Re: Getting RH8 to recognize new drive on cciss controller

Stuart Browne — Wed, 17 Sep 2003 06:25:24 GMT

Ok, it doesn't seem like you've got much of an understanding of RAID container's.

First off, details of your existing containers. What sort of container is currently configured for your existing disks?

You mention RAID1, but not how many disks are in the container. If it's just 2, you should be able to fool the new machine into thinking it's the master of a pair who's lost it's sibling, if you are careful, and get the RAID controller to think tha the disk has the correct container information, and not the controller.. (read: power machine off, clear controller's config, plug disk in)

As for the other, you've put a new disk in. You've create a new container for the new disk, or are you trying to join it on to the existing container?

To reconfigure/add container details to a running system, you need to access the ACUXE (which should have been one of the packages you installed). It's accessed via a browser, pointing to port 2301 (https). The documentation that comes with the RAID controller details what the interface looks like, and how to use it.

There are no 'shell' utilities that I'm aware of to do this sort of thing.

Good luck!

Re: Getting RH8 to recognize new drive on cciss controller

Johan Sunnerstig — Thu, 18 Sep 2003 11:10:35 GMT

Thanks for the reply.

The main problem is that I can't mount the disk in Linux.
I've been into ACUXE and setup a logical volume with only the new disk in it, and it looks fine from both ACUXE and the management tools.

So, the setup on the hardware side should be done, no?
The problem is just that I can't find it inside Linux.

The short question is, how can I make RH detect an added disk(logical volume)?
Without restarting that is.

As for the other question, it's not really important, I've done similar things, moving single disks from 2 disk RAID1's to another empty box, then making it rebuild the array with another fresh disk.
Just wondering if something similar can be done, except when putting the disk in a storage enclosure rather than in another machine.

In this particular case, it was merely an ugly way to move a load of data to another site, the data is on a DL360 with two 36 GB disks in a RAID1.
Would it be possible to pull one of the disks, put it in the enclosure, then have the controller just pick it up as a broken RAID1?

Regards
Johan

Re: Getting RH8 to recognize new drive on cciss controller

Stuart Browne — Thu, 18 Sep 2003 22:19:05 GMT

The answer to the later comment should be Yes. And I agree, it's an ugly way to do it ;)

Now, getting linux to find a new logical volume on a cciss..

What's currently listed in /proc/driver/cciss/cciss* ? Is it listing both volumes?

Is there a /dev/cciss/c1d0 currently? If there isn't, it might use the first available logical volume number. It seems odd that the first volume is appearing as the 2nd volume.

device numbering should be:

c0d0 = 1st volume on 1st controller
c0d1 = 2nd volume on 1st controller

thus

c1d1 = 2nd volume on 2nd controller

Anyway.. Have a look at those details in the cciss* files.

Re: Getting RH8 to recognize new drive on cciss controller

Stuart Browne — Thu, 18 Sep 2003 22:51:27 GMT

Oh, side note with regards to the 'rescan'.. That's only for systems which are using hot-swap tapes, and have had the 'engage scsi' sent to the controller device.

According to the documentation, it won't do anything with regards to logical volumes..