Dell R610 hardware raid 1 configuration
 

This is a long story, so please bear with me - I am trying to give as much information as possible. I have a Dell R610 and an MD1000 powervault, which is to be used entirely for data storage. The R610 arrived first, so I thought I'd set about installing the OS, CENTOS 5.3. The machine is equipped with twin 146 GB disks. In the PERC configuration, I set it to RAID 1, and then in the CENTOS setup I manually repartitioned the disk with separate partitions for /var, /tmp, /xlog, /boot, /user and / . Install proceeded without a hitch and the machine worked beautifully. Next, the Powervault arrived and I set about configuring it. We have three disks in RAID 0, and four disks in RAID 5 and a hotspare for the RAID 5 partition. After having setup the RAID information in the PERC, I went on to set up the Linux partitions on the LVM, which also went without incident. But I was surprised to see the LVM reporting that there were unitialised partitions on /dev/sde. Further investigations showed that /dev/sde5 is the bootable partition on the R610. The uninitialised partitions are reported as sde1-sde4, and they contain the original partitions as shipped by Dell.

I confess that I was puzzled as to why the bootable device was now a partition on /dev/sde. Of course, I didn't check which /dev device it was before, so maybe it was that all the time, but I'd be surprised.
All seemed to work, however, and when the machine is running I can see the activity lights on both the R610 disks coming on in sync most of the time - I assumed that this is during write processes to mirror the disks as planned. I also assumed that when the disk 0 was lit alone, it was reading and that when disk 1 was lit alone it was somehow having to catch up with disk 0.

This morning I decided to test the mirroring and I deliberately pulled disk 0, and I got a message that a disk was missing and that I could press any key to continue, which I did. I was relieved that the machine came up without a hitch. So I powered the machine off, put disk 0 back in and removed disk 1 and then powered On again. The OS did not boot, and got I messages saying that I could retry the boot, import a foreign configuration and other things that I didn't think were a good idea. I powered the machine off again and replaced disk 0, but on reboot I got all the same trouble - it wouldn't boot and it would try to boot from the NIC. I checked the boot sequence, which was as it should be - CDROM, hard disk, NIC. Finally, I examined the BIOS boot device and found that somehow the boot RAID controller had switched from the embedded controller to the additional on for the SAS powervault. I changed those around and retried - the machine booted without any problem, but the error light on disk 0 flashed constantly while that on disk 1 was on solid. The activity lights on both disks were also flashing very rapidly. However, eventually they went off and the error light on disk 0 came on solid.

After all that I thought I'd have a look at the LVM and I see that /dev/sde is still reporting uninitialised partitions. But /dev/sde5 is also reported in the volume group as the boot device with my original partitions. I have obviously not done everything I need to do to set up mirroring, but I can't figure out what it is - can someone please help me? I'd be very grateful!

Thanks,

Karel

First, congrats on getting all that hardware! It took me 6 months from purchase request to receiving to get my R610s.

Upon putting back Disk 0, did you rebuild your RAID? It's a simple step, but since you were so detailed, it seems out of place that you didn't mention the biggest part of the process. Upon booting, either from the BIOS, RAID BIOS, or Dell Open Manage software, you need to rebuild your virtual disk. As far as I've experienced, Dell hardware doesn't automatically resync your RAID.

Firstly, huge thanks for taking the time to have an interest in helping people with their computer problems. I appreciate it very much.

I confess that it didn't occur to me that I'd need to rebuild a mirrored configuration - I knew I'd probably have to do that for RAID 5, but in my simplistic view of RAID 1 I thought that all data was written to both disks simultaneously and that one disk was an exact copy of the other. I also thought that this was confirmed when I removed disk 0 and the machine was perfect on disk 1 alone, so imagine my horror to discover that it didn't happen when disk 0 was driving the machine on its own. I went into a kernel panic all of my own! At any rate, I now know that even RAID 1 needs to be rebuilt (although I'm not sure how it gets rebuilt - I'll look into that.)

Thanks again for taking the time to read my post and especially for an informative reply.

BTW - I'm sure the supply chain is easier now. I had heard that Dell was having problems shipping units some time ago. My two boxes came within a week!