It turns out the cause was missing 63-md-raid-arrays.rules.  It
appears 64-md-raid-assembly.rules will assemble the devices on
boot/sysinit and works even without 63-md-raid-arrays.rules, but
starting using the script needs the 63-md-raid-array.rules.  Not sure
why, but that appears to be it.   If I remove 64 and leave 63, the
dmraid takes over, but I can "dmsetup remove_all" then start the md
RAID with the script and it works, I can also use nodmraid and then
run the script.

So thanks for pointing me to udev - still would be curious why 64
doesn't need 63.

On Tue, Aug 16, 2022 at 10:54 PM Hannes Reinecke [off-list ref] wrote:

On 8/17/22 04:04, David F. wrote:

quoted

What rules should be used?   I don't see a /dev/md directory, I
created one, stopped the raid (all the /dev/md* devices went away)
and tried to start the raid, same thing and only /dev/md127 gets
created, nothing in /dev/md/ directory and none of the md126 devices ?
  You then get the timeout.

Please check the device-mapper status.

My guess is that device-mapper gets in the way (as it probably will be
activated from udev, too), and blocks the devices when mdadm is trying
to access them.

Cheers,

Hannes
--
Dr. Hannes Reinecke                Kernel Storage Architect
hare@suse.de                              +49 911 74053 688
SUSE Software Solutions GmbH, Maxfeldstr. 5, 90409 Nürnberg
HRB 36809 (AG Nürnberg), Geschäftsführer: Ivo Totev, Andrew
Myers, Andrew McDonald, Martje Boudien Moerman