Glossary of terminologies

Service interruption: Any event that makes part of a component's services unavailable, including but not limited to client APIs, web UIs, or admin interfaces.

Rolling Upgrade - Process of upgrading all the components without any service interruptions. Zero Downtime upgrade and upgrade by rolling restart are synonyms of Rolling Upgrade.

Rolling Restart - Ability to restart a service without any service interruption to a component or dependent components.

High Availability (HA) - The ability to avoid a service interruption when individual components fail within a service.

Quick Restart - Component restart with a service interruption but with minimal downtime.

Worker Roles - Roles that are related to data storage and processing. For example, HDFS has Datanode and Impala has Impala Daemon, and so on.

Non-Worker - Roles that are related to master, admin, coordinator, or non-data-related activities. For example, HDFS-Namenode, Journalnode Impala- StateStore, Catalog, Hue’s non-worker roles, such as the load balancer, Kerberos ticket renewer, and Hue server, and so on.

Downgrade - Downgrade restores the software to the prior release version but data remains intact. Service interruptions may or may not be required for the downgrade process, depending on the component.

Rollback - Rollback the state of the data of the upgraded components to the pre-upgrade state. A Rollback may require a service interruption and is not an automated operation.