Upgrade a cluster's operating system

Upgrade a Redis Enterprise Software cluster's operating system to a later major version.

To upgrade the operating system (OS) on a Redis Enterprise Software cluster to a later major version, perform a rolling upgrade. Because you upgrade one node at a time, you can upgrade your cluster's OS without downtime.

Prerequisites

Before you upgrade a cluster's operating system:

1. Upgrade all nodes in the cluster to a Redis Enterprise Software version that supports the OS's current version and upgrade version.

   To learn which versions of Redis Enterprise Software support specific OS versions, see Supported platforms.

2. If the cluster uses custom directories, make sure the OS upgrade version also supports custom directories, and specify the same custom directories during installation for all nodes. See Customize installation directories for details.

Perform OS rolling upgrade

To upgrade the cluster's operating system, use one of the following rolling upgrade methods:

• Extra node method - recommended if you have additional resources available

• Replace node method - recommended if you cannot temporarily allocate additional resources

Extra node upgrade method

1. Complete all prerequisites before starting the rolling upgrade.

2. Create a node with the OS upgrade version.

3. Install the cluster's current Redis Enterprise Software version on the new node using the installation package for the OS upgrade version.

4. Add the new node to the cluster.

5. If the cluster uses DNS, add the new node’s IP address to the DNS records.

6. Remove one node running the earlier OS version from the cluster.

7. Repeat the previous steps until all nodes with the earlier OS version are removed. If the final node to remove from the cluster is the primary node, demote it to a secondary node before you remove it.

Replace node upgrade method

1. Complete all prerequisites before starting the rolling upgrade.

2. Remove a node with the earlier OS version from the cluster.

3. Uninstall Redis Enterprise Software from the removed node:

   sudo ./rl_uninstall.sh
   

4. Either upgrade the existing node to the OS upgrade version, or create a new node with the OS upgrade version.

5. Install the cluster's current Redis Enterprise Software version on the upgraded node using the installation package for the OS upgrade version.

6. Add the new node to the cluster.

   If you want to reuse the removed node's ID when you add the node to the cluster, run rladmin cluster join with the replace_node flag:

   rladmin cluster join nodes <cluster_member_ip_address> username <username> password <password> replace_node <node_id>
   

7. If the cluster uses DNS, add the new node’s IP address to the DNS records.

8. Verify node health:

   1. Run rlcheck on all nodes:

      rlcheck
      

      The output lists the result of each verification test:

      ##### Welcome to Redis Enterprise Cluster settings verification utility ####
      Running test: verify_bootstrap_status
                      PASS
      ...
      Running test: verify_encrypted_gossip
                      PASS
      Summary:
      -------
      ALL TESTS PASSED.
      

      For healthy nodes, the expected output is ALL TESTS PASSED.

   2. Run rladmin status on the new node:

      rladmin status extra all
      

      The expected output is the OK status for the cluster, nodes, endpoints, and shards:

      CLUSTER:
      OK. Cluster master: 2 (<IP.address>)
      Cluster health: OK, [0, 0.0, 0.0]
      failures/minute - avg1 0.00, avg15 0.00, avg60 0.00.
      ...
      

9. Repeat the previous steps until all nodes with the earlier OS version are replaced. If the final node to remove from the cluster is the primary node, demote it to a secondary node before you remove it.