Node Replacement

PCS: Node Replacement

Guildline

• <node-name> mean pacemaker cluster node Eg: srvnode-1 srvnode-2
• <faulty node> is Node which is gone bad
• <new node> is new node replaced with <faulty node>
• <working node> is node which is part of cluster other that <faulty node>

Configure Pacemaker Cluster

• PCS Cluster Setup

Replace node

• Poweroff faulty node
  • Put node on standby if node is still running
  • If node is already poweroff then ignore this step

$ pcs cluster standby <faulty-node>
$ poweroff

Remove faulty node from cluster

• To add identical node to cluster we need to remove old node

$ pcs cluster node remove <faulty-node>

Update hosts on nodes [new and working]

• Update /etc/hosts

<ip1>   srvnode-1
<ip2>   srvnode-2

Configure New Node for Pacmaker setup

• Follow Step-[1 to 3] From PCS Cluster Setup
• Authorize new node from working node

$ pcs cluster auth srvnode-1 srvnode-2

• Add node to cluster (Execute from <working node>)

$ pcs cluster node add `<new node>`

Enable and start node for cluster

• After Enable and start node
  • All resource will get sycn on nodes
  • All Rules are get sync on node
  • <new node> will become part of cluster and replace <faulty node>
• Enable and start node

$ pcs resource cleanup --all # To clean history
$ pcs cluster enable <new-node>
$ pcs cluster start <new-node>