Can't RaidStart /dev/md3 in Raid1 configuration during attempted rebuild.
The short of it (don't ask questions why I did things a certain way, it was a long day and I wasn't thinking clearly).
Red Hat 7.3
Software Raid 1 mirroring sda to sdb, md0 through md5.
Recently we had a cooling system failure and sda was the victim.
An identical drive was installed to rebuild from sdb to sda.
I STOPPED /dev/md3 and unmounted it (/boot) This was in a misunderstanding on how to rebuild the array.
I realized this was not necessary and went about it this way:
I formatted the new sda to match the current sdb.
raidhotadd /dev/mdx /dev/sdx for watched as the md's rebuilt themselves. It was glorious.
However, I still needed to rebuild /dev/md3 (mirroring partitions sda1 and sdb1).
Obviously, I cannot raidhotadd a device to md3 if md3 is not ACTIVE!
raidstart /dev/md3 return "invalid argument."
/var/log/messages shows the following:
md: invalid raid superblock magic on sda1
md: sda1 has invalid sb, not importing!
md: could not import sda1!
md: autostart sda1 failed!
Now. This technically makes sense, as sda1 has not been rebuilt from sdb1! Thus, it can't start the raid device because part of it is degraded or not present (it is, afterall, just a formatted partition). Nevertheless, sdb1 DOES work. So it seems I merely can't start the raid /dev/md3 because it wants to start it with both sda1 and sdb1 active and working... which it obviously can't do.
SO, the question is this: How can I force the raid device /dev/md3 to start in degraded mode using ONLY /dev/sdb1? If I could do that, I'd just raidhotadd /dev/md3 /dev/sda1 and watch the magic. I'd like to do this without rebooting the system, as this is the head node to a cluster of ours that people like to use alot.
Can anyone point me in the right direction?