>Unless there is something on the primary that you need to keep, just re-sync
>from the active partition and next time they will come up paired:
>mdadm --add md_device missing_partition
>cat /proc/mdstat to see the status
Yeah, a one-off test was done to the secondary which was not desired to keep.
So the primary was needed to be brought back as current.
>I think there is a count of clean shutdowns that is used to
>determine if one is more current than the other - maybe there is a more
>intelligent way to update it, though.
I'd love to find this out
I ended up pruning the array while booted off the secondary, booting from the
primary solo and pruning its array, then rebooting off the primary with both
and extending the array and adding the secondary to it. Sigh, that was a PITA.
CentOS mailing list