A rolling restart of a DSE cluster is an important task that must be performed after some significant change has been made to a cluster and is documented in various processes.
A rolling restart is not a full cluster down, then cluster back up process but rather the downing of a single node and restart of the DSE process, this allows us to leave a cluster operational while performing changes to a cluster one node at a time; this is a very important uptime process unique to DSE/Cassandra.
TerraDSE contains a playbook that automates this process, you can invoke it on an ad-hoc basis by the following command:
ansible-playbook dse_cluster_rolling_restart.yml
Please note that if you are using:
- A manual inventory (explicit "hosts" file) that you will need to acertain it's a true representation of your cluster before running this command.
- If using a dynamic inventory please acertain the accuracy of your TFSTATE file/s are a true representation of your cluster.