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

Re: RAID b/w GPT and NON GPT partition.



On 01/07/15 07:01 PM, Arno Schuring wrote:

Date: Wed, 1 Jul 2015 18:41:35 -0400
From: garydale@torfree.net

On 01/07/15 03:24 PM, Don Armstrong wrote:
On Wed, 01 Jul 2015, Gary Dale wrote:
The size of the RAID array is set by the smallest partition so if you
want to be able to boot from either drive, then putting the ef02
partition in the free space on the new drive means that you will
either not be able to boot from the old drive should the new drive
fail, or you will have different partition tables on each drive.
This is precisely why you should have the EF02 partition on every single
drive in the raid set and run grub-install on all of them (or at the
very least, one more than the number of drives that can fail and still
assemble the array).
You missed the point that this would require different partition tables
on the two drives.
Whose point was that? It certainly wasn't the OP's, who asked to have a
GPT partition table on the second disk of a raid array, whilst the
first had MBR.
So we should always do things that will likely lead to confusion or mistakes later on?
Moreover, if you ever replaced the original drive
with a larger one, you would have to install LVM to get around the ef02
partition
This makes no sense. There is no "around" if you created the ef02
partition in the first available sectors of the disk, as everyone has
been advocating. Nor would it make sense to "work around" a partition
of less than 1MB, you would only need to "work around" it if you made
it unsensibly large, like the 100MB you seem to favour. And why would
you want identical partition tables across drives of different sizes
in the first place?
This was in response to your suggestion that you put the ef02 partition in extra space on the new drive. Your suggestion, not mine. I wanted the partitions to be the same on both drives with the exception of the extra space on the new drive - which I would normally assign to the last RAID array so it can be grown if the smaller driver gets replaced.



[LVM] an added layer of complexity that you shouldn't need.
Sure. Because no one would ever need to resize or migrate volumes
without downtime. Don't patronize people by telling them what they do
or do not need. We have seen more than enough of that already.
That's a rather specific requirement that most people don't need. Adding complexity for a feature you don't need is not a good idea. It's just one more thing that can go wrong.


However you seem to have conceded the larger point about UEFI which M$
is trying to make mandatory.
The way I read the thread is that people have been trying to correct
your misconceptions, not argue with you. But meh, I'm sure I'm reading
it wrong.
Do you deny that M$ has been pressuring manufacturers to make UEFI mandatory or that it would require a larger ef02 partition than you have been promoting as the right way to do it? Most distros have been working toward existing in a UEFI world. And since we're already in a TB to PB world, why fret over things that are so insignificant as 100M? If this was the 1980s, you'd have a point but we're three decade past that.


Reply to: