Upgrade each Hive Metastore server one at a time across all Hive Metastore nodes. There is no order that needs to be followed. Before upgrading Hive, back up the database as described in "Prepare the Cluster."
Note: Existing connections to the Metastore will be dropped during this process, and there is a chance that running CLI jobs may fail. However, the Metastore will automatically retry any failed jobs.
Find the Hive Metastore server process ID and stop the process. For example:
# ps aux | grep hive ... hive 129414 0.1 3.3 2002400 131332 ? . . . org.apache.hadoop.hive.metastore.HiveMetaStore ... # kill 129414
Switch to the new software version, and start the server:
hdp-select set hive-metastore 2.2.6.0-2800
su - hive -c "nohup /usr/hdp/current/hive-metastore/bin/hive --service metastore>/var/log/hive/hive.out 2>/var/log/hive/hive.log &"