Pacemaker : Configure GFS2 Filesystem

Set GFS2 Filesystem Resource on Cluster. This example is based on the environment like follows.   1) Basic Cluster setting is done   2) Fence Device is configured +--------------------+ | [ ISCSI Target ] | | dlp.srv.world | +---------+----------+ 10.0.0.30| | +----------------------+ | +----------------------+ | [ Cluster Node#1 ] |10.0.0.51 | 10.0.0.52| [ Cluster Node#2 ] | | node01.srv.world +----------+----------+ node02.srv.world | | | | | +----------------------+ +----------------------+

[1]	Create a shared storage on ISCSI Target for GFS2 Filesystem, refer to here. On this example, it created ISCSI storage as IQN [iqn.2021-06.world.srv:dlp.target02] with [10G] size.
[2]	On all Cluster Nodes, Install required packages and change LVM settings.

# enable [HighAvailability, ResilientStorage] repo and install (disabled by default) [root@node01 ~]#  dnf --enablerepo=highavailability,resilientstorage -y install lvm2-lockd gfs2-utils dlm [root@node01 ~]#  vi /etc/lvm/lvm.conf # line 1172 : uncomment and change use_lvmlockd =  1

[3]	On a Node in cluster, configure DLM resource for GFS2 filesystem.

# set [no-quorum-policy=freeze] on GFS2 [root@node01 ~]#  pcs property set no-quorum-policy=freeze # create controld resource # [dlm] ⇒ any name you like # [group] ⇒ any group name [root@node01 ~]#  pcs resource create dlm ocf:pacemaker:controld op monitor interval=30s on-fail=fence group locking --future # create clone of [locking] to activate it on all nodes in cluster [root@node01 ~]#  pcs resource clone locking interleave=true # create lvmlockd resource # [lvmlockdd] ⇒ any name # [group] ⇒ the same group with controld resource [root@node01 ~]#  pcs resource create lvmlockdd ocf:heartbeat:lvmlockd op monitor interval=30s on-fail=fence group locking --future # verify status # OK if all [Started] [root@node01 ~]#  pcs status --full Cluster name: ha_cluster Cluster Summary: * Stack: corosync (Pacemaker is running) * Current DC: node02.srv.world (2) (version 2.1.6-10.1.el9-6fdc9deea29) - partition with quorum * Last updated: Tue Dec 5 12:45:00 2023 on node01.srv.world * Last change: Tue Dec 5 12:44:49 2023 by root via cibadmin on node01.srv.world * 2 nodes configured * 6 resource instances configured Node List: * Node node01.srv.world (1): online, feature set 3.17.4 * Node node02.srv.world (2): online, feature set 3.17.4 Full List of Resources: * scsi-shooter (stonith:fence_scsi): Started node01.srv.world * Resource Group: ha_group: * lvm_ha (ocf:heartbeat:LVM-activate): Started node01.srv.world * Clone Set: locking-clone [locking]: * Resource Group: locking:0: * dlm (ocf:pacemaker:controld): Started node01.srv.world * lvmlockdd (ocf:heartbeat:lvmlockd): Started node01.srv.world * Resource Group: locking:1: * dlm (ocf:pacemaker:controld): Started node02.srv.world * lvmlockdd (ocf:heartbeat:lvmlockd): Started node02.srv.world Migration Summary: Fencing History: * unfencing of node01.srv.world successful: delegate=node01.srv.world, client=pacemaker-controld.4616, origin=node02.srv.world, completed='2023-12-05 11:15:32.582894 +09:00' * unfencing of node02.srv.world successful: delegate=node02.srv.world, client=pacemaker-fenced.4606, origin=node01.srv.world, completed='2023-12-05 11:15:32.564894 +09:00' Tickets: PCSD Status: node01.srv.world: Online node02.srv.world: Online Daemon Status: corosync: active/enabled pacemaker: active/enabled pcsd: active/enabled

[4]	On a Node in cluster, configure shared volume on shared storage. [sdb] on the example below is shared storage from ISCSI Target.

# discover target [root@node01 ~]#  iscsiadm -m discovery -t sendtargets -p 10.0.0.30 10.0.0.30:3260,1 iqn.2022-01.world.srv:dlp.target01 10.0.0.30:3260,1 iqn.2022-01.world.srv:dlp.target02 # login [root@node01 ~]#  iscsiadm -m node --login --target iqn.2022-01.world.srv:dlp.target02 [root@node01 ~]#  iscsiadm -m session -o show tcp: [1] 10.0.0.30:3260,1 iqn.2022-01.world.srv:dlp.target01 (non-flash) tcp: [2] 10.0.0.30:3260,1 iqn.2022-01.world.srv:dlp.target02 (non-flash) # set LVM [root@node01 ~]#  parted --script /dev/sdb "mklabel gpt" [root@node01 ~]#  parted --script /dev/sdb "mkpart primary 0% 100%" [root@node01 ~]#  parted --script /dev/sdb "set 1 lvm on" # create physical volume [root@node01 ~]#  pvcreate /dev/sdb1 Physical volume "/dev/sdb1" successfully created. # create chared volume [root@node01 ~]#  vgcreate --shared vg_gfs2 /dev/sdb1 Volume group "vg_gfs2" successfully created VG vg_gfs2 starting dlm lockspace Starting locking. Waiting until locks are ready...

[5]	Move to another node and start lock manager for shared volume.

[root@node02 ~]#  iscsiadm -m discovery -t sendtargets -p 10.0.0.30 10.0.0.30:3260,1 iqn.2022-01.world.srv:dlp.target01 10.0.0.30:3260,1 iqn.2022-01.world.srv:dlp.target02 [root@node02 ~]#  iscsiadm -m node --login --target iqn.2022-01.world.srv:dlp.target02 [root@node02 ~]#  iscsiadm -m session -o show tcp: [1] 10.0.0.30:3260,1 iqn.2022-01.world.srv:dlp.target01 (non-flash) tcp: [2] 10.0.0.30:3260,1 iqn.2022-01.world.srv:dlp.target02 (non-flash) [root@node02 ~]#  lvmdevices --adddev /dev/sdb1 [root@node02 ~]#  vgchange --lock-start vg_gfs2 VG vg_gfs2 starting dlm lockspace Starting locking. Waiting until locks are ready... [root@node02 ~]#  vgs VG #PV #LV #SN Attr VSize VFree cs 1 2 0 wz--n- <29.00g 0 vg_gfs2 1 0 0 wz--ns <9.98g <9.98g

[6]	Back to the node you set shared volume and create logical volume and configure GFS2 filesystem.

# create logical volume [root@node01 ~]#  lvcreate -l 100%FREE -n lv_gfs2 vg_gfs2 Logical volume "lv_gfs2" created. # format with GFS2 [root@node01 ~]#  mkfs.gfs2 -j2 -p lock_dlm -t ha_cluster:gfs2-01 /dev/vg_gfs2/lv_gfs2 This will destroy any data on /dev/dm-3 Are you sure you want to proceed? [y/n] y Discarding device contents (may take a while on large devices): Done Adding journals: Done Building resource groups: Done Creating quota file: Done Writing superblock and syncing: Done Device: /dev/vg_gfs2/lv_gfs2 Block size: 4096 Device size: 9.98 GB (2615296 blocks) Filesystem size: 9.98 GB (2615293 blocks) Journals: 2 Journal size: 32MB Resource groups: 42 Locking protocol: "lock_dlm" Lock table: "ha_cluster:gfs2-01" UUID: 1f5ed3d1-4785-4ecd-bc02-02c282bac54f # create LVM-activate resource # [shared_lv] ⇒ any name # [group] ⇒ any group name [root@node01 ~]#  pcs resource create shared_lv ocf:heartbeat:LVM-activate lvname=lv_gfs2 vgname=vg_gfs2 activation_mode=shared vg_access_mode=lvmlockd group shared_vg --future # create clone of [LVM-activate] [root@node01 ~]#  pcs resource clone shared_vg interleave=true # set start order as [locking] → [shared_vg] [root@node01 ~]#  pcs constraint order start locking-clone then shared_vg-clone Adding locking-clone shared_vg-clone (kind: Mandatory) (Options: first-action=start then-action=start) # set that [shared_vg] and [locking] start on a same node [root@node01 ~]#  pcs constraint colocation add shared_vg-clone with locking-clone # create Filesystem resource # [shared_fs] ⇒ any name # [device] ⇒ device with GFS2 formatted # [directory] ⇒ any directory you'd like to mount GFS2 filesystem # [group] ⇒ the same group with LVM-activate resource [root@node01 ~]#  pcs resource create shared_fs ocf:heartbeat:Filesystem device="/dev/vg_gfs2/lv_gfs2" directory="/home/gfs2-share" fstype="gfs2" options=noatime op monitor interval=10s on-fail=fence group shared_vg --future # verify status # OK if all [Started] [root@node01 ~]#  pcs status --full Cluster name: ha_cluster Cluster Summary: * Stack: corosync (Pacemaker is running) * Current DC: node02.srv.world (2) (version 2.1.6-10.1.el9-6fdc9deea29) - partition with quorum * Last updated: Tue Dec 5 13:11:43 2023 on node01.srv.world * Last change: Tue Dec 5 13:11:19 2023 by root via cibadmin on node01.srv.world * 2 nodes configured * 10 resource instances configured Node List: * Node node01.srv.world (1): online, feature set 3.17.4 * Node node02.srv.world (2): online, feature set 3.17.4 Full List of Resources: * scsi-shooter (stonith:fence_scsi): Started node01.srv.world * Resource Group: ha_group: * lvm_ha (ocf:heartbeat:LVM-activate): Started node01.srv.world * Clone Set: locking-clone [locking]: * Resource Group: locking:0: * dlm (ocf:pacemaker:controld): Started node01.srv.world * lvmlockdd (ocf:heartbeat:lvmlockd): Started node01.srv.world * Resource Group: locking:1: * dlm (ocf:pacemaker:controld): Started node02.srv.world * lvmlockdd (ocf:heartbeat:lvmlockd): Started node02.srv.world * Clone Set: shared_vg-clone [shared_vg]: * Resource Group: shared_vg:0: * shared_lv (ocf:heartbeat:LVM-activate): Started node01.srv.world * shared_fs (ocf:heartbeat:Filesystem): Started node01.srv.world * Resource Group: shared_vg:1: * shared_lv (ocf:heartbeat:LVM-activate): Started node02.srv.world * shared_fs (ocf:heartbeat:Filesystem): Started node02.srv.world Migration Summary: Fencing History: * unfencing of node01.srv.world successful: delegate=node01.srv.world, client=pacemaker-controld.4616, origin=node02.srv.world, completed='2023-12-05 11:15:32.582894 +09:00' * unfencing of node02.srv.world successful: delegate=node02.srv.world, client=pacemaker-fenced.4606, origin=node01.srv.world, completed='2023-12-05 11:15:32.564894 +09:00' Tickets: PCSD Status: node01.srv.world: Online node02.srv.world: Online Daemon Status: corosync: active/enabled pacemaker: active/enabled pcsd: active/enabled # OK if GFS2 filesystem is mounted on both nodes [root@node01 ~]#  df -hT /home/gfs2-share Filesystem Type Size Used Avail Use% Mounted on /dev/mapper/vg_gfs2-lv_gfs2 gfs2 10G 67M 10G 1% /home/gfs2-share [root@node02 ~]#  df -hT /home/gfs2-share Filesystem Type Size Used Avail Use% Mounted on /dev/mapper/vg_gfs2-lv_gfs2 gfs2 10G 67M 10G 1% /home/gfs2-share