[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Software RAID



Mark McCulligh wrote:
Hi Group,

I have software raid running on a server. Two harddrives on one ribbon.
Primary and Secondary. I setup software raid to mirror between the two
drives.

My question is if the Primary drive fails, then the secondary drive will
kick in.

Possibly, but there's a possibilty that a dead drive will make that IDE channel inoperative and render the slave useless. It's much better to have the 2 drives on separate channels, not just for the above reason but also for speed as having both disks in a RAID on the same channel will be slower.


But what happens if I reboot the machine without a working Primary
drive. Will it boot or will I have to first open the case remove the primary
drive and make the secondary drive primary before it will boot. Then add a
new Secondary drive and rebuild the raid array.  OR can I just replace the
primary drive with a new drive, the server booting from the secondary and
then rebuilds the raid array on the primary.


That depends on how you've configured your drives and which boot manager you use. Is your boot partition mirrored, and do you use lilo or grub? If the secondary doesn't have a boot partition then it won't be able to boot. If it has a boot partition which is a mirror of the primary, and you use grub it won't be able to boot as the secondary, you'd have to swap it to primary, but if you use lilo then it should boot as the secondary. lilo is RAID aware and if your boot partition is mirrored it writes a different boot sector to each device in the RAID so it is bootable in it's current location.


I am trying to figure out the recovery method from a failed primary drive.
No good to have raid if you don't know to use it.

With grub you would have to swap drives, insert a new secondary and rebuild the RAID. With lilo you just insert a new primary, boot off the secondary, partition the new disk and rebuild the raid. At least, that's how it worked when I was testing my firewall...


There's no substitute for testing. Try it out. Remove the primary and see if the system boots. Remove the partition table from the primary and see what steps are necessary in rebuilding the RAID. It's all good experience.
When I was building my firewall this is exactly what I did to see how the system handled a failed disk, and how I should go about adding a replacement. I ended up writing a Python script which would partition a new, blank, replacement disk and add it into the raid. I'm currently building a new mail server with mirrored disks and will do exactly the same again with this system.


Thanks,
Mark.



--
Nigel Wade, System Administrator, Space Plasma Physics Group,
            University of Leicester, Leicester, LE1 7RH, UK
E-mail :    nmw ion le ac uk
Phone :     +44 (0)116 2523548, Fax : +44 (0)116 2523555




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]