The short story is I think I need to add drives back that were removed from a raid array by accident.

My power supply started failing but didn't just shut down. The computer was running but the drives spun down (yikes!). This made the raid software, I assume, think the drives failed and removed them from the array.

The problem is, too many were removed from one of the partitions (naturally the one I need the most). I replaced the power supply and all drives spin up and are recognized in the bios.

On boot I get:

Warning: /dev/disk/by-uuid/105f2e75-6900-42d1-85c5-8d9ab70da32a does not exist
I also get a notice that md126 does not have enough operational mirrors
then ...
Entering emergency mode
then ...
dracut:/#


mdstat has:
md125 : active (auto-read-only) raid10 sda5[0] sdc5[2]
md126 : inactive sda3[0]
md127 : active (auto-read-only) raid10 sda2[0] sdc2[2]


Tried mdadm --assemble --scan
I got:
mdadm: /dev/md/localhost.localdomain:databases assembled from drive - not enough to start the array
mdadm: /dev/md/localhost.localdomain:root_0 assembled from drive - not enough to start the array
mdadm: /dev/md/localhost.localdomain:swap_0 assembled from drive - not enough to start the array
mdadm: /dev/md/localhost.localdomain:home assembled from drive - not enough to start the array

Tried mdadm --assemble --scan /dev/md126
I got:
mdadm : /dev/md126 not identified in config file

I ran mdadm --detail on each of the above. md125 and md127 each have raid drives 1 and 3 missing (the array has 4 total). md126 has raid drives 1,2 & 3 missing.
md125 shows /dev/sda5 and /dev/sdc5 are there. I'm guessing I could run mdadm --manage /dev/md125 --add /dev/sdb5 --add sdd5 to put them back in
md126 shows /dev/sda3 as there. I'm guessing I could run mdadm --manage /dev/md126 --add /dev/sdb3 --add sdc3 --add sdd3 to put them back in
md127 shows /dev/sda2 and /dev/sdc3 are there. I'm guessing I could run mdadm --manage /dev/md127 --add /dev/sdb2 --add /dev/sdd2 to put them back in


Since I've never done this before, I'm looking for some input from folks who have experience with this.

Always show the actual output, instead of telling us what the output looked like...

Run both `mdadm --detail /dev/md*` and `mdadm --examine /dev/sd*` and copy-paste the whole thing in a [ code ] block.

mdadm --add usually only works when the RAID itself is still running.