Full Restart Upgrade

The preferred way to upgrade a running CrateDB cluster is via the Rolling Upgrade method. But this is not always an option (for example, when you are upgrading from one feature version to the next).

When a Rolling Upgrade is not possible, a full restart upgrade must be done if you want to update CrateDB.

Caution

These instructions work for most ways of installing and running CrateDB. However, if you are using Docker, take note that docker service update performs a Rolling Upgrade.

To do a full restart upgrade with Docker, you must remove the CrateDB service and then recreate it.

Upgrade Process

Warning

Before upgrading, you should back up your data.

Stop Every Node

Once you have backed up your data and made the necessary preparations for downtime, stop the crate daemon on every node in the cluster, using whatever method is appropriate to your setup.

Upgrade Every Node

Warning

Make sure that every crate daemon has been stopped before you continue or you risk data corruption or data loss.

Once the cluster has been stopped, upgrade the CrateDB software on every node in the cluster, using whatever method is appropriate to your setup.

Start Every Node

Once the CrateDB software on node in the cluster has been updated, start the crate daemon on every node in the cluster, using whatever method is appropriate to your setup.