-
Committer:
Ryan Harper
-
Date:
2017-05-03 14:06:56 UTC
-
mfrom:
(437.2.28 trunk.md-uefi)
-
Revision ID:
ryan.harper@canonical.com-20170503140656-we42q1jgwc2clpa1
clear-holders: mdadm use /proc/mdstat to wait for array to stop
shutdown_mdadm needs to wait until the md device has stopped. Using
files in sysfs is unreliable due to a kernel bug (LP:1682456) so instead
use device presence in /proc/mdstat.
Add the use of --manage flag to force mdadm to interrupt actions on the
device (like a resync).
Add vmtest with storage config to recreate issue (LP:1682584) found using
early_command to dirty existing disks with raid configurations.
Fixes: LP: #1672584