FreeBSD Adaptec 2100S RAID
Date: 02 April 2005
Setting up the 2100S is actually pretty simple. Here’s how I did it.
Configure your Kernel
The 2100S uses the asr driver which is included in the GENERIC kernel, however, I take that out of my custom kernels since most of my machine do not have a RAID controller at all, so I added this line to my kernel config and rebuilt.
device asr # DPT SmartRAID V, VI and Adaptec SCSI RAID
Install the Adaptec RAID Control Software
The CD that comes with the controller includes some software to control and
configure the RAID. What bugs me is that, be default, it want’s to install
itself into /usr/dpt and I’m not really thrilled with installing third-party
software in /usr. That’s OK though, it’s easy enough to install it in /usr/
local by using the -p option to pkg_add.
mount /cdrom
pkg_add -p /usr/local/dpt /cdrom/package/FreeBSD/apps/sm/adptfbsd_314.tgz
umount /cdrom
Configure the RAID
Now is the time to stop and make sure that you have your HDDs installed and ready. The drives need to be pristine, i.e. they cannot have any existing partitions, etc. and cannot have the same SCSI ID as the controller. (The 2100S uses ID 7 by default.)
It’s probably a good idea to disable kern.securelevel before proceeding as well, as it may prevent the controller from accessing the drives properly.
Now that everything is ready, it’s time to create the array. Let’s pretend that we have eight HDDs with IDs 1-6, 8 and 9 (remember, the controller is using ID 7), and we want to build a RAID-10 array.
Note: I used raidutil to do the setup because I don’t have X installed on that server. Feel free to use Storage Manager or Storage Manager Pro.
Identify the Controller Address
We need the Adaptec address for the controller we’re using for the RAID. In my case, the ID is d0. The ID will be used to tell raidutil which controller we’re creating the RAID on.
bash-2.05b# ./raidutil -L controller
# b0 b1 b2 Controller Cache FW NVRAM Serial Status
---------------------------------------------------------------------------
d0 -- ADAP2100S 16MB 370F CHNL 1.1 BBXXXXXXXXXOptimal
Get the Adaptec Addresses for the HDDs
raidutil requires that you refer to your HDDs by their Adaptec addresses. You
can get the the HDD addresses with the physical parameter to -L. Note: -c d0
is what tells raidutil which controller to look at. Remember, d0 is the ID we
got above.
bash-2.05b# ./raidutil -c d0 -L physical
Address Type Manufacturer/Model Capacity Status
---------------------------------------------------------------------------
d0b0t1d0 Disk Drive (DASD) IBM IC35L018UCPR15-0 17501MB Optimal
d0b0t2d0 Disk Drive (DASD) IBM IC35L018UCPR15-0 17501MB Optimal
d0b0t3d0 Disk Drive (DASD) IBM IC35L018UCPR15-0 17501MB Optimal
d0b0t4d0 Disk Drive (DASD) IBM IC35L018UCPR15-0 17501MB Optimal
d0b0t5d0 Disk Drive (DASD) IBM IC35L018UCPR15-0 17501MB Optimal
d0b0t6d0 Disk Drive (DASD) IBM IC35L018UCPR15-0 17501MB Optimal
d0b0t8d0 Disk Drive (DASD) IBM IC35L018UCPR15-0 17501MB Optimal
d0b0t9d0 Disk Drive (DASD) IBM IC35L018UCPR15-0 17501MB Optimal
Build the RAID-10
Now it’s time to build the RAID. The setup below creates the stripe across mirrored pairs. The + between addresses denotes the mirroring. You can mirror the drives in any order. What seemed a little weird to me was setting the RAID level (-l) to 1 but it works.
Note: The raidutil man page that came on my CD seems to have an error when discussing RAID-10 configutation. From the man page:
To create a RAID 10 configuration specify groups in pairs as follows:
raidutil -c c1 -l 1 -g c1t1d0,c1t2d0,c1t3d0,c1t4d0
Notice that there are no + signs anywhere in that command to denote drive pairs.
Note: -g must be the last option to the command. If it is not, your other options will not be applied.
./raidutil -c d0 -l 1 -z 64 -g d0b0t1d0+d0b0t2d0,d0b0t3d0+d0b0t4d0,d0b0t5d0+d0b0t6d0,d0b0t8d0+d0b0t9d0
You can use -L raid to watch the controller build the RAID. When it’s done, you should see something like this.
bash-2.05b# ./raidutil -c d0 -L raid
Address Type Manufacturer/Model Capacity Status
---------------------------------------------------------------------------
d0b0t1d0 RAID 0 (Striped) ADAPTEC RAID-10 70007MB Optimal
d0b0t1d0 RAID 1 (Mirrored) ADAPTEC RAID-1 17501MB Optimal
d0b0t1d0 Disk Drive (DASD) IBM IC35L018UCPR15-17501MB Optimal
d0b0t2d0 Disk Drive (DASD) IBM IC35L018UCPR15-17501MB Optimal
d0b0t3d0 RAID 1 (Mirrored) ADAPTEC RAID-1 17501MB Optimal
d0b0t3d0 Disk Drive (DASD) IBM IC35L018UCPR15-17501MB Optimal
d0b0t4d0 Disk Drive (DASD) IBM IC35L018UCPR15-17501MB Optimal
d0b0t5d0 RAID 1 (Mirrored) ADAPTEC RAID-1 17501MB Optimal
d0b0t5d0 Disk Drive (DASD) IBM IC35L018UCPR15-17501MB Optimal
d0b0t6d0 Disk Drive (DASD) IBM IC35L018UCPR15-17501MB Optimal
d0b0t8d0 RAID 1 (Mirrored) ADAPTEC RAID-1 17501MB Optimal
d0b0t8d0 Disk Drive (DASD) IBM IC35L018UCPR15-17501MB Optimal
d0b0t9d0 Disk Drive (DASD) IBM IC35L018UCPR15-17501MB Optimal
You’re Done
Congratulations! You now have a working RAID. The RAID should be available as
/dev/da0. You can now label and partition the RAID as you would for any other
drive.
Replacing a Failed Drive
I recently came in to the tune of the RAID controller’s alarm going off. The server just kept chugging way without a hitch. I’m beginning to like RAID.
cd /usr/local/dpt
root@accounts dpt# ./raidutil -A
# b0 b1 b2 Controller Alarm Status Alarm Enable
---------------------------------------------------------------------------
d0 -- 2100S On Enabled
root@accounts dpt# ./raidutil -L physical | less
Address Type Manufacturer/Model Capacity Status
---------------------------------------------------------------------------
d0b0t1d0 Disk Drive (DASD) IBM IC35L018UCPR15-0 17501MB Failed drive
d0b0t2d0 Disk Drive (DASD) IBM IC35L018UCPR15-0 17501MB Optimal
d0b0t3d0 Disk Drive (DASD) IBM IC35L018UCPR15-0 17501MB Optimal
d0b0t4d0 Disk Drive (DASD) IBM IC35L018UCPR15-0 17501MB Optimal
d0b0t5d0 Disk Drive (DASD) IBM IC35L018UCPR15-0 17501MB Optimal
d0b0t6d0 Disk Drive (DASD) IBM IC35L018UCPR15-0 17501MB Optimal
d0b0t8d0 Disk Drive (DASD) IBM IC35L018UCPR15-0 17501MB Failed drive
d0b0t9d0 Disk Drive (DASD) IBM IC35L018UCPR15-0 17501MB Optimal
It looks like I lost two drives. That’s not good at all. However, this is still survivable. Replace the drives then run:
./raidutil -c d0 -a rebuild d0b0t1d0
Now when you run raidutil -L raid you should be something like this:
root@accounts dpt# ./raidutil -c d0 -L raid
Address Type Manufacturer/Model Capacity Status
---------------------------------------------------------------------------
d0b0t1d0 RAID 0 (Striped) ADAPTEC RAID-10 70007MB Reconstruct 6%
d0b0t1d0 RAID 1 (Mirrored) ADAPTEC RAID-1 17501MB Reconstruct
d0b0t1d0 Disk Drive (DASD) IBM IC35L018UCPR15-17501MB Replaced Drive
d0b0t2d0 Disk Drive (DASD) IBM IC35L018UCPR15-17501MB Optimal
d0b0t3d0 RAID 1 (Mirrored) ADAPTEC RAID-1 17501MB Reconstruct
d0b0t3d0 Disk Drive (DASD) IBM IC35L018UCPR15-17501MB Optimal
d0b0t4d0 Disk Drive (DASD) IBM IC35L018UCPR15-17501MB Optimal
d0b0t5d0 RAID 1 (Mirrored) ADAPTEC RAID-1 17501MB Reconstruct
d0b0t5d0 Disk Drive (DASD) IBM IC35L018UCPR15-17501MB Optimal
d0b0t6d0 Disk Drive (DASD) IBM IC35L018UCPR15-17501MB Optimal
d0b0t8d0 RAID 1 (Mirrored) ADAPTEC RAID-1 17501MB Reconstruct
d0b0t8d0 Disk Drive (DASD) IBM IC35L018UCPR15-17501MB Replaced Drive
d0b0t9d0 Disk Drive (DASD) IBM IC35L018UCPR15-17501MB Optimal
This will eventually finish and you should be good to go again.