If you deployed multiple VMware Identity Manager virtual appliances in a cluster, you must stop the RabbitMQ cluster on all nodes before you upgrade the VMware Identity Manager appliance.
About this task
The RabbitMQ nodes must be stopped in the reverse order that they were started. This preserves the order of the master node. To determine the start order, view the /db/rabbitmq/data/*/nodes_running_at_shutdown files on each server. Shut down the RabbitMQ node that lists all the nodes first. For example, if you have three nodes that were started as node1, then node2, then node3, the nodes_running_at_shutdown file on node 3 lists node1,node2,node3. Node 2 lists node1,node2. Node 1 list node1. You shut down node 3, then node 2, then node 1.
- Stop RabbitMQ nodes on each VMware Identity Manager appliance in the cluster. Type rabbitmqctl stop.
Do this for each RabbiMQ node in the cluster before continuing.
- Verify that RabbitMQ is detached from the cluster. Type rabbitmqctl cluster_status.
- Upgrade the first node. See the upgrade procedures either in Upgrading VMware Identity Manager Online or Upgrading VMware Identity Manager Offline.
The VMware Identity Manager appliance is started.
- Follow steps 2 through 4 for each node.
As each node is upgraded, run the
rabbitmgctl cluster_statuscommand on the upgraded node to verify that all the nodes upgraded so far are listed in the
running_nodessection of the output. After upgrading node 1, the
running_nodessection lists only node1. After upgrading node 2, run the
rabbitmqctl cluster_statuscommand on both nodes and the
running_nodessection should each list node1 and node2. This indicates that the RabbitMQ nodes are clustered together correctly.
When all nodes are upgraded, RabbitMQ forms a cluster with the nodes in the correct order.