On 2018-01-15 13:36, Anton Aylward wrote:
I just realized...
I use LVM. It lets me do two things in this arena.
The fist, obviously, is that I can take a snapshot of a file system. if might happen that the LE I take that onto is on the new drive :-) That way I don't have to work around UUID duplication and a plethora of other things.
The second: well I can tell LVM that I have extended LE to the second drive, then tell LVM to migrate all the sectors in that LE from the first drive to the second, then deallocate the the ones on the first drive. As far as LVM is concerned there was one FS and hence on UUDI that for part of the time was distributed over two drives (and how is that different from RAID?). At the beginning it was all on the first drive; at the end it was all on the second drive.
At different times I've done both of these. Oh, and you can do them on live file systems. I did the latter once, while working with a DBA to optimize a 200T DB2 database on a large IBM AIX installation. All while the database was running and serving customers. We then said OK to the IBM techs who came in and carted out a bank of RAID drives that we'd just freed up on trolleys and carted in and commissioned and new bank, which we then introduced into the LVM array and redistributed and balanced the database and it performed better of course. We made a point of not telling the users any of this: why worry them?
It may seem like a lot of work when written out here, but it's all an excellent example of making the machine work while you go get a coffee (something I'm in favour of), reliable and uncomplicated. LVM is a wonderful tool.
While that is true, the use case here was a backup copy for restoring in case of disaster, so it needs to be to a different and separate media. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" (Minas Tirith))