Andreas Enge <andreas@enge.fr> writes:

> On Sat, Jul 30, 2016 at 07:05:25PM -0400, myglc2 wrote:
>> > I might write a more detailed blog post about this; there is a little
>> > subtlety with the non-automatic determination of dependencies between
>> > devices, so one needs to make sure that the partitions to be assembled
>> > are present before the mdadm command is executed.
>> I rebooted and the array is not assembled ;-(
>
> Strange! But I will also tell you the subtlety ;-)  Here is a trick to use
> (thanks to Ludovic):
> (define md0
>   (mapped-device
>     (source (list "/dev/sda2" "/dev/sdb2"))
>       (target "/dev/md0")
>       (type raid-device-mapping)))
> (operating-system
> ...
>   (mapped-devices (list md0))
>   (file-systems (cons (file-system
>                         (title 'device)
>                         (device "/dev/md0")
>                         (dependencies (list md0))
>                         (mount-point "/")
>                         (type "ext4"))
>                       %base-file-systems))
>
> The "dependencies" field makes sure that the file system is only mounted
> after the array is assembled; I am not sure that this is your problem,
> but you might want to give it a try.
>
> In the long run, this should be reprogrammed: Devices and file systems
> should wait until all their "inputs" are present, or at least wait for
> a reasonable time.

Thanks, I tried that. the 'guix system reconfigure' succeeds and starts
the raid array (please see system35.scm & system35.log, attached).

But the reboot hangs at:

[...] clocksource: Switched to clocksource tsc