Hello Neil,

On 06/25/2014 03:03 AM, NeilBrown wrote:
[...]

quoted

After waiting a couple of min, stopping again md126 worked:

[ 1835.755661] md: md126 stopped.

Is this expected ?

No overly surprising.

This is probably caused by udev, or something udev runs, opening /dev/md126
after it has been stopped.  This has the effect of creating an empty inactive
array.
e.g.

Why opening /dev/md126 after it's been stopped didn't fail ?

Also why doing 'mdadm --stop /dev/md126' several times reports that the
array was stopped each time ?

 mknod /dev/test b 9 57
 < /dev/test

will make /sys/block/md57 appear.

It is a bit untidy, but shouldn't cause problems.

Another 'inconsistent' state I noticed is when a new array is created.
There's a time window when /sys//block/md127/md exists but the contained
disk (dev-sda1 dev-sdb1, for example) are not present in that directory.
This state (as the previous one) confuses my tool that is listening udev
events in order to register all block devices present in the system.

So I am wondering how I should deal with this 'transition' states and
how can detect them in a reliable way ?

Thanks.