How to Add and Remove Nodes in a Redis Cluster

Scalability is a key metric for software systems, and horizontal scaling is essential for achieving it. Redis Cluster provides a simple way to add or remove nodes, enhancing scalability. The following sections illustrate these operations with concrete examples.

Adding Nodes

There are two scenarios for adding a node to an existing cluster: as a master or as a slave.

1. Add as Master Create a new directory (e.g., cluster-test/7006) and copy a redis.conf file, then modify key settings such as:

port 7006                     # port number
cluster-enabled yes          # enable cluster mode
cluster-config-file nodes-7006.conf  # cluster state file
cluster-node-timeout 5000    # timeout in ms
dbfilename "dump-7006.rdb"   # data file
protected-mode no
logfile "logfile-7006.log"
requirepass 123456
masterauth 123456
pidfile "redis_7006.pid"

Start the instance on port 7006 and add it to the cluster with:

redis-cli --cluster add-node 127.0.0.1:7006 127.0.0.1:7000 -a 123456

The command’s first argument is the new node’s address, the second is any existing cluster node, and -a supplies the password. Verify the addition with cluster nodes, which shows the new node as a master without slots or election participation.

2. Add as Slave Use the same add-node command but include --cluster-slave:

redis-cli --cluster add-node 127.0.0.1:7006 127.0.0.1:7000 --cluster-slave -a 123456

This randomly selects a master with few replicas. To attach the new node to a specific master (e.g., master 7002), run the command with the target master’s ID, after which the node appears as a slave of that master.

Removing Nodes

To remove a node, use del-node:

redis-cli --cluster del-node 127.0.0.1:7000 723d06ef31c5ada88ee3926249fc89f69fae3767 -a 123456

The first argument is any cluster node, the second is the node ID to delete. If the node is a master, ensure it holds no data before removal.

For a failed node that cannot be contacted, del-node will fail. Instead, execute cluster forget on all remaining nodes:

redis-cli --cluster call 127.0.0.1:7000 cluster forget <node-id>

This forces every node to forget the dead node’s ID.

Conclusion

The examples show that adding or removing nodes in a Redis cluster is straightforward, providing an effective horizontal scaling solution for high‑availability data stores.