Re: RAID1 with multiple partitions
On 10.12.2010 21:15, David Gaudine wrote:
The system works, but I still have some questions about whether what
I've already done is OK.
1) Is it OK to use 4 MD arrays like this, or should I use just one and
LVM like the jerryweb link? I got a bit confused reading about LVM
and I don't need anything fancy like being able to resize partitions.
I personally prefer LVM as it gives me more flexibility, but if you do
not need that, the setup you selected is fine.
2) I put the SWAP partition on RAID. The first guide doesn't use RAID
for swap. The author emailed me his comments about the pros and cons,
and I think I want it on RAID for peace of mind. It shouldn't really
matter since I have much more RAM than I need. Is there any reason I
might regret putting SWAP on RAID? "cat /proc/mdstat" reports the MD1
(the swap device) as "auto-read-only".
In Linux, all raid arrays stay in "auto-read-only" mode until they are
access the first time after each reboot, so this seems to imply that
your system has not yet initialized the swap. Did you add your swap to
/etc/fstab and mark it as swap there?
3) The first guide doesn't use a separate boot partion, the second
does. Comments? I've never used a separate boot partition. I think
it used to be important on large disks, to keep the kernel in the
first 1024 cylinders, but isn't important anymore.
A separate boot partition still makes sense for example if you would
like to use a root file system that is not directly supported by your
boot manager. I am not sure if the 1024 cylinders border is still valid.
Maybe somebody else can comment on this?
4) The first guide shows how to install Grub on both disks. After
that's done once, do I have to do it again whenever there's a new
kernel package? Or in any other situation that I have to watch out for?
You only need to install grub on both disks once. After that, there is
no need to repeat that again unless you upgrade grub itself.
For the other two questions, I cannot say anything.
Hope this helps.
Regards,
Reiner.
Reply to: