hi,

here about a tips to add / remove an LV using raid mirror with multiple slice and all that with HA-cluster RHEL 7.1

HA-cluster is monitored by Pacemaker

why this ticket ?

Just becaus there is a redhat bug implementation and RedHat support don't find alone solution.

So, first the need is : (we have already 4 PV using multipath, and 1 VG "vgtest" already using pacemaker resource)

lvcreate -m1 -i2 -I 128 -L 200m -n test4 vgtest /dev/mapper/pv1-1 /dev/mapper/pv2-1 /dev/mapper/pv1-2 /dev/mapper/pv2-2


First, we have to stop resource onto active node
pcs resource disable $resource_name_vgtest

then we need to reconfigure /etc/lvm/lvm.conf

locking_type = 1
ignore_lvm_mirrors = 0 => (set 1 if ou need remove a failed or inactive  LV/VG)
use_lvmetad = 1
## Add the VG into locale volume list group:
volume_list = [ "vgroot", "vgtest" ]

reload lvm :
systemctl restart lvm2-lvmetad

# create your LV with raid 1 onto 4 slice #
lvcreate -m1 -i2 -I 128 -L 200m -n test4 vgtest /dev/mapper/pv1-1 /dev/mapper/pv2-1 /dev/mapper/pv1-2 /dev/mapper/pv2-2
# format it - default xfs under RHEL 7 #
mkfs.xfs /dev/vgtest/test4  or /dev/mapper/vgtest-test4


## backoff local LV to cluster ##
edit /etc/lvm/lvm.conf and change to this properties
 volume_list = [ "vgroot" ]

# reload lvm service
systemctl restart lvm2-lvmetad

# enable pacemaker resource
pcs resource enable $resource_name_vgtest


You can check it :
pcs resource  || pcs status
## all have to be "started" ##

## Then you can mount this Lv as any other FileSystem mount method ##

## If not all started, use a clean to purge all failed and redetect status
pcs resource cleanup $Resource_name