CentOS - Jan 2019 - C7 and mdadm - apparently SOLVED

Looking around on the Web, what I just tried was to stop the RAID, then do
an --assemble --scan, and *that* found everything, put the raid together,
and appears to be rebuilding using the new drive.

*phew*

I do find it interesting that scan works, but explicitly assembling
apparently remembers the drive that died, and wouldn't even look at it.

     mark

mark wrote:
> Looking around on the Web, what I just tried was to stop the RAID, then
> do an --assemble --scan, and *that* found everything, put the raid
> together, and appears to be rebuilding using the new drive.
>
> *phew*
>
>
> I do find it interesting that scan works, but explicitly assembling
> apparently remembers the drive that died, and wouldn't even look at it.
>
The rebuild said 6000+ min, then seemed to stop. I see errors on one
drive, but can't seem to do anything. It had /dev/md0 up, but insisted
there were two spares. I tried to get them added, and now it won't start
up /dev/md0.

    mark

Does anyone know if the actual physical position in the hot-swap bays
affects how mdadm tries to assemble a drive?

     mark