On Sat, May 24, 2014 at 11:39 AM, Mikael Abrahamsson [off-list ref] wrote:

On Sat, 24 May 2014, Mikael Abrahamsson wrote:

quoted

Your output seems to indicate that you have one drive with an event count
that is 6 events off the others. This most likely means it's safe to use
--assemble --force in order to make this get accepted.

I am able through some combinations to get mdadm to load "an" array
("an" because I'm pretty sure its using a blank disk), however, I get
errors when trying to mount the filesystem (I have to force the type
(xfs) afterwhich it errors out with invalid superblock).
I have up until this point assumed the array was being built
incorrectly. Is it possible that the filesystem is simply just
corrupt?

Actually, it seems you have two drives with event counts that are off, one
more than the other. Make sure you only add the drives with the highest
event count, then add the last drive later so it gets sync:ed from the other
drives. The event count isn't that much off (I believe it was 40 events)
even on the most differing one, so even with that your filesystem shouldn't
be too much at risk.

Until you have everything under control I wouldn't write the filesystem
read/write though, to avoid further problems...


--
Mikael Abrahamsson    email: swmike@swm.pp.se