...
- Edit ams-site, Set timeline.metrics.service.operation.mode = distributed
- Edit ams-hbase-site,
- Set hbase.rootdir = hdfs://<namenode-host>:8020/user/ams/hbase [ If NN HA is enabled, hdfs://<nameservice-id>/user/ams/hbase ]
(Note: /user/ams/hbase here is the directory where metric data will be stored in HDFS) - Set hbase.cluster.distributed = true
- Add dfs.client.read.shortcircuit = true (This is an optimization with a local DN present)
- Set hbase.rootdir = hdfs://<namenode-host>:8020/user/ams/hbase [ If NN HA is enabled, hdfs://<nameservice-id>/user/ams/hbase ]
- In Ambari versions > 2.2.0 make sure hbase.zookeeper.property.clientPort=2181 (cluster Zookeeper port).
- Restart Metrics Collector
Note: In Ambari 2.0.x, there is a bug in deploying AMS in distributed mode, if Namenode HA is enabled. Please follow the instruction listed in this JIRA as workaround steps: (AMBARI-10707)
Note: In Ambari 2.2.1, stack advisor changes the dependent configs for distributed mode automatically through recommendations. Ideally, the only config that needs to be changed is timeline.metrics.service.operation.mode = distributed. The other configs - hbase.rootdir, hbase.cluster.distributed and dfs.client.read.shortcircuit will be changed automatically.
Migrating data from embedded to distributed mode
...