Differences

This shows you the differences between two versions of the page.

--- software:cluster:rhcs_pacemaker_rhel8 [2021/09/15 04:34] – chris
+++ software:cluster:rhcs_pacemaker_rhel8 [2022/11/13 12:06] (current) – external edit 127.0.0.1
@@ Line 1: / Line 1: @@
+Setup of a simple RHEL8 cluster with pacemaker.
+  * install 2 nodes with rhel8.4
+  * https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/8/html-single/configuring_and_managing_high_availability_clusters/index
+<code>
+# following to be done on both nodes
+subscription-manager register
+subscription-manager list --all --available >allsubs.txt
+less allsubs.txt
+# choose right pool with HA
+subscription-manager subscribe --pool=<poolid>
+subscription-manager repos --disable='*' \
+        --enable=rhel-8-for-x86_64-baseos-rpms \
+        --enable=rhel-8-for-x86_64-appstream-rpms \
+        --enable=rhel-8-for-x86_64-highavailability-rpms \
+        --enable=rhel-8-for-x86_64-resilientstorage-rpms
+# ensure /etc/hosts on both nodes has entries of both nodes
+echo '192.168.4.23    node2.local node2' >>/etc/hosts
+echo '192.168.4.22    node1.local node1' >>/etc/hosts
+yum install -y pcs pacemaker fence-agents-all
+systemctl enable --now pcsd
+echo -n redhat | passwd hacluster --stdin
+# execute only on first node
+pcs host auth node1.local node2.local
+pcs cluster setup my_cluster --start node1.local node2.local
+systemctl enable --now pcsd corosync pacemaker
+### fencing
+# setup hypervisor as per
+# https://access.redhat.com/solutions/917833
+# https://access.redhat.com/solutions/55920
+pcs stonith list
+# At this point, we should have file /etc/cluster/fence_xvm.key
+# on both nodes.
+[root@node1 ~]# fence_xvm -o list
+ahub                             75d52bd3-a5cb-429e-94e6-47c509c8f109 off
+atower                           e5692f79-1f68-4a62-96a4-37b3bfde2561 off
+fedora34c                        bd1538fa-b051-49b5-9f07-b7a57788fa8b off
+node1                            87ebc6c6-3015-43a6-9c49-00044d8f3d7c on
+node2                            a40a6534-4a8e-4e58-9f69-5e2b9b81615e on
+rhel7.9                          4bba326d-08dc-4289-aa12-7f9f6c9f6a3e off
+rhel8.4                          daf830c0-c947-4373-a10e-12e3b6c3fa84 off
+# https://www.golinuxcloud.com/what-is-fencing-configure-kvm-cluster-fencing/
+pcs stonith create fence-node1 fence_xvm port=node1 pcmk_host_list=node1.local
+pcs stonith create fence-node2 fence_xvm port=node2 pcmk_host_list=node2.local
+pcs stonith status
+# shared disk, iscsi from hypervisor
+# as per https://fluxcoil.net/wiki/software/iscsi/target
+# ensure /etc/iscsi/initiatorname.iscsi on your nodes is different!
+md5sum /etc/iscsi/initi*
+systemctl enable --now iscsi
+iscsiadm --mode discovery --type sendtargets --portal 192.168.4.1
+iscsiadm -m node --target iqn.2003-01.org.x:disk1 -p 192.168.4.1 -l
+# I have no other volume groups on both nodes, so have
+# this in /etc/lvm/lvm.conf on both nodes:
+auto_activation_volume_list = []
+system_id_source = "uname"
+# at this point, both nodes should see a common /dev/sda
+# create a single partition, just to follow best practices:
+parted /dev/sda 'mklabel gpt'
+parted /dev/sda 'mkpart 1 2048s -1'
+pvcreate /dev/sda1
+vgcreate httpd_vg /dev/sda1
+lvcreate -L450 -n httpd_lv httpd_vg
+mkfs.ext4 /dev/httpd_vg/httpd_lv
+vgs --noheadings -o vg_name  # should show the VG just on one node
+# then, rebuild initrd and reboot
+dracut -H -f /boot/initramfs-$(uname -r).img $(uname -r)
+reboot
+# after coming up, cluster should have started already
+# pcs cluster start --all
+pcs status
+# This should show the nodename now in the VG:
+vgs -o+systemid
+# create LVM cluster resource
+pcs resource create my_lvm ocf:heartbeat:LVM-activate \
+  vgname=httpd_vg vg_access_mode=system_id --group apachegroup
+pcs resource status
+pcs status # gives complete status
+pcs resource move my_lvm node1.local # moves resource to node1
+</code>

fluxcoil wiki

User Tools

Site Tools

Differences

Page Tools