Description of problem:

CI jobs running on the bare-metal lab occasionally fail to provision one or more workers because of software RAID on one of the disks.

The block device looks like this:

NAME="md127" KNAME="md127" PATH="/dev/md127" MAJ_MIN="9:127" FSAVAIL="" FSSIZE="" FSTYPE="" FSUSED="" FSUSE_PCT="" FSROOTS="" FSVER="" MOUNTPOINT="" MOUNTPOINTS="" LABEL="" UUID="" PTUUID="" PTTYPE="" PARTTYPE="" PARTTYPENAME="" PARTLABEL="" PARTUUID="" PARTFLAGS="" RA="128" RO="0" RM="0" HOTPLUG="0" MODEL="" SERIAL="" SIZE="239307063296" STATE="" OWNER="root" GROUP="disk" MODE="brw-rw----" ALIGNMENT="0" MIN_IO="4096" OPT_IO="0" PHY_SEC="4096" LOG_SEC="512" ROTA="1" SCHED="" RQ_SIZE="" TYPE="raid1" DISC_ALN="0" DISC_GRAN="4096" DISC_MAX="0" DISC_ZERO="0" WSAME="0" WWN="" RAND="0" PKNAME="sda1" HCTL="" TRAN="" SUBSYSTEMS="block" REV="" VENDOR="" ZONED="none" DAX="0"

If a member holder device is picked as a root one, openshift-installer (rightfully) fails:

coreos-installer: Partitions in use on /dev/sda:
coreos-installer: /dev/sda1 in use by /dev/md127
coreos-installer: Error: checking for exclusive access to /dev/sda

Example of such a run: https://prow.ci.openshift.org/view/gs/test-platform-results/logs/periodic-ci-openshift-release-master-nightly-4.19-e2e-metal-ipi-ovn-bm-upgrade/1918809704964820992