You can change the location of the socket files anytime, but to limit downtime, it is best to perform this change when you are installing or upgrading Redis Enterprise Software to version 4.5.0.-51 or higher. If this change needs to be performed immediately on an existing cluster, see the instructions below.

Changing the Location on an Existing Redis Enterprise Software (RS) Installation

You must be running RS version 4.5.0-51 or higher, and with the sock files location is still under the default location of /tmp. (This is where they go when is executed without -s flag.)

  1. Execute this command on all nodes:

    $ sudo rlutil create_socket_path socket_path=/var/run/redislabs
  2. Execute this command on the master node:

    $ sudo rlutil set_socket_path socket_path=/var/run/redislabs
  3. Execute this command on all nodes in a serial manner:

    $ sudo service rlec_supervisor restart
  4. There are two ways to do the next step depending on your uptime requirements.

    1. Restart each database in the cluster, but you will incur downtime. This is by far the easiest option to execute.

      $ rladmin restart db <db name>
    2. Execute the following three commands for each master/slave shards during a maintenance window. All three of these commands must be executed on the node hosting the shard being operated on:

      $ redis_ctl stop <slave shard id>
      $ rladmin failover shard <master shard id>
      $ redis_ctl stop <new slave shard id>


      $ redis_ctl stop 6 # command run on node 1, since shard 6 is running on node 1
      $ rladmin failover shard 5
      $ redis_ctl stop 5 # command run on node 3, since shard 5 is running on node 3

      This must be performed for each database in the cluster and each shard in each database.