On 27/11/2020 23:10, c.buhtz@xxxxxxxxx wrote:
Hello,
I red some stuff about assembling mode for mdadm. But two points are
unclear for me.
1. Missing devices
I have two HDDs in a RAID1 - only data, no OS, on them. I want to
physical remove one HDD and plug it into a new server machine. Then I
want to bring the RAID1 online again on the new machine but with one
existing and one missing drive.
Background: When finished I will plug in a new fresh HDD as the second
one to the new server.
It's fairly certain that the array will - initially - refuse to run.
That's a safety feature - if an array is damaged while shut down, it
won't come back without intervention.
Put the single disk in your new system. As I say, it's unlikely it'll
start straight away. "cat /proc/mdstat" and if the array is there in a
failed state, stop it. Then you can re-assemble it and it should come up
degraded. All you have to do after that is replace the non-existent
drive with your new replacement, and you'll have a functioning array.
2. mdadm.conf
On the web I read sometimes about modifying mdadm.conf on the new
machine. But I do not understand why. If so. Why and what do I have to
modify in the mdadm.conf?
If you've even got one ...
I don't know the history of it, but the early arrays did not have
superblocks, so mdadm.conf - an ACCURATE - mdadm.conf was essential. Now
they've got superblocks, it's optional and - to the best of my knowledge
- none of my systems have mdadm.conf's.
Do you have a superblock? Preferably v1 (1.0, 1.1 or 1.2). If you do,
don't worry about it, it's old advice, and while it's good to have an
up-to-date mdadm.conf to tell *you* what's what, the system doesn't need it.
Cheers,
Wol
