What is the process for upgrading an HA cluster?

JFrog Support
2016-10-06 13:38

Upgrading an HA cluster version older than 3.4.0 will require that you shut down all HA nodes, perform all required upgrades on your primary (master) and secondary (slave) nodes, wait 5 minutes and then bring all of everything up again. 

For Artifactory 3.4.0 and above, improvements to the upgrade process have been implemented.  You should have zero downtime during your upgrade if you follow the process below:

    • We will assume that you are working with a cluster running with one primary and any number of secondary nodes.

    • Bring down your primary node and perform its upgrade; the load balancer will redirect all queries to the secondary nodes while you are doing this.

    • When the primary node starts up again, it will update the configuration and the DB schema, and then send a reload configuration message to the secondary nodes.

    • After receiving the reload-message from the primary node, all secondary nodes will compare their version to the version running on the primary node; if the versions do not match they will shut down gracefully.

    • Run the upgrade procedure on all secondary nodes and bring them back up to bring the whole cluster up to the latest version.