Creating Pacemaker Cluster Resources

6. Creating resources¶

Virtual IP¶

pcs resource create vip ocf:heartbeat:IPaddr2 ip=192.168.13.34 cidr_netmask=24 \
    op monitor interval=20s

Gotcha: don't put the subnet in the ip= parameter — ip="192.168.0.200/24" is wrong and fails with 'not configured' (6) / [findif] failed. Use ip= for the bare address and cidr_netmask= separately:

pcs resource create etcd_ip ocf:heartbeat:IPaddr2 ip="192.168.0.200" cidr_netmask=24 \
    op monitor interval=60s

HAProxy (systemd resource, cloned across all nodes)¶

pcs resource create haproxy systemd:haproxy clone

Apache¶

pcs resource create httpd ocf:heartbeat:apache \
    configfile=/etc/httpd/conf/httpd.conf \
    statusurl="http://127.0.0.1/server-status" \
    op monitor interval=20s

Needs mod_status enabled and a <Location /server-status> block allowing 127.0.0.1 in the Apache config on every node.

MariaDB/MySQL (simple, non-Galera)¶

pcs resource create mariadb ocf:heartbeat:mysql op monitor interval=1min

Galera (multistate master/master cluster)¶

pcs resource create galera ocf:heartbeat:galera \
    enable_creation=true \
    wsrep_cluster_address="gcomm://node1,node2,node3" \
    meta master-max=3 ordered=true \
    op monitor interval=20s timeout=30s promote timeout=300s

Colocation / ordering / location constraints¶

pcs constraint colocation add vip httpd mariadb INFINITY
pcs constraint order start vip then start httpd
pcs constraint location vip prefers node1=50

Grouping resources (start/stop together, in order, on the same node)¶

pcs resource create cluster_ip0 ocf:heartbeat:IPaddr2 ip=192.168.122.70 \
    cidr_netmask=24 op monitor interval=30s --group public_ips
pcs resource create cluster_ip1 ocf:heartbeat:IPaddr2 ip=192.168.122.71 \
    cidr_netmask=24 op monitor interval=30s --group public_ips

Staged config editing pattern (batch several changes atomically)¶

Useful for anything with several interdependent resources (DRBD + filesystem + service, for example) — build the change set offline against a snapshot of the CIB, review it, then push it live in one shot.

pcs cluster cib work.cfg                      # snapshot current live CIB
pcs -f work.cfg resource create ... 
pcs -f work.cfg constraint ...
pcs -f work.cfg resource show                 # review before committing
pcs cluster cib-push work.cfg                 # or: pcs cluster push cib work.cfg (older pcs)

← Part 03: Fencing & STONITH · Part 05: Troubleshooting →