| 1 |
On Mon, 24 Jun 2019 18:00:29 +0100, Peter Humphrey wrote: |
| 2 |
|
| 3 |
> > I feel like device-mapper should be its own package that other things |
| 4 |
> > depend on; LVM, RAID (mdadm, et al.), multi-path, LUKS (cryptsetup). |
| 5 |
> |
| 6 |
> Yes, and that would fit the Unix tradition of doing one thing, and |
| 7 |
> well. It should also be applied to crypt-setup and its friends. |
| 8 |
|
| 9 |
Indeed, all it needs is someone willing to do the work. Now if LVM had |
| 10 |
been written by Lennart Poettering, there would be a stampede of |
| 11 |
volunteers to do this, witness eudev. |
| 12 |
|
| 13 |
> > > On this box, which does need lvm for RAID-1 on two SSDs: |
| 14 |
> > Do you /need/ LVM? Or is it extra that comes with device-mapper? |
| 15 |
> |
| 16 |
> No, I do actually use lvm to base a raid-1 file system on. I haven't |
| 17 |
> considered raid-1 without lvm; is that feasible? |
| 18 |
|
| 19 |
Of course it is, a RAID1 device is just a block device on which you can |
| 20 |
put any filesystem you like. RAID and LVM are complementary technologies |
| 21 |
that work well together, but neither needs the others (apart from the |
| 22 |
device-mapper bit). |
| 23 |
|
| 24 |
|
| 25 |
-- |
| 26 |
Neil Bothwick |
| 27 |
|
| 28 |
The road to HAL is paved with good intentions. |
Archives