Upgrade non-core cluster components
These components are master-only, no slave/worker processes.
Upgrade non-core cluster components one at a time across all nodes, in the following order:
Upgrade Hive
Upgrade Oozie
Upgrade Falcon
- 1. Overview
- 2. Preparation
- 3. Upgrading the Cluster
- Upgrade ZooKeeper
- Upgrade Ranger
- Upgrade core-cluster master processes
- Upgrade core slave nodes (HDFS, YARN, HBase)
- Upgrade non-core cluster components
- Upgrade Knox
- Upgrade components outside of the cluster
- Upgrade Slider
- Upgrade clients on cluster (core and non-core) components
- Finalize the rolling upgrade process
- 4. Downgrading the Cluster
- Downgrade clients on cluster components
- Downgrade Slider
- Downgrade components outside of the cluster
- Downgrade Knox
- Downgrade non-core cluster components
- Downgrade core slave nodes
- Downgrade core-cluster master processes
- Downgrade the HBase Thrift server
- Downgrade HBase REST Server
- Downgrade HBase Master(s)
- Downgrade the YARN ResourceManager process or HA process pair
- Downgrade the YARN Timeline Service
- Downgrade the MapReduce JobHistoryServer
- Downgrade the HDFS NameNode HA Process Pair
- Downgrade HDFS JournalNode(s)
- Validation
- Downgrade Ranger
- Downgrade ZooKeeper
- 5. Initiating Rollback
- 6. Appendix A
- 7. Appendix B