@mike_bronson7  I think the only change you will need to make is the "KAFKA" service in Uppercase because "kafka" in lowercase will not exist as a service. Following is the example which i tested in my cluster and it works pretty fine. # export service=kafka (INCORRECT) # export service=KAFKA # curl -iLv -u "admin:admin" -i -H 'X-Requested-By: ambari' -X PUT -d '{"RequestInfo":{"context":"_PARSE_.STOP.$service","operation_level":{"level":"SERVICE","cluster_name":"$CLUSTER_NAME","service_name":"$service"}},"Body":{"ServiceInfo":{"state":"INSTALLED"}}}' http://$HOST:8080/api/v1/clusters/$CLUSTER_NAME/services/$service   In order to start KAFKA service. # curl -iLv -u "admin:admin" -i -H 'X-Requested-By: ambari' -X PUT -d '{"RequestInfo":{"context":"_PARSE_.START.$service","operation_level":{"level":"SERVICE","cluster_name":"$CLUSTER_NAME","service_name":"$service"}},"Body":{"ServiceInfo":{"state":"STARTED"}}}' http://$HOST:8080/api/v1/clusters/$CLUSTER_NAME/services/$service   .   ... View more

@mike_bronson7    As we see the error as: "Table Namespace Manager not fully initialized" 2020-02-21 03:33:49,284 INFO org.apache.hadoop.hbase.client.RpcRetryingCaller: Call exception, tries=15, retries=35, started=629725 ms ago, cancelled=false, msg=java.io.IOException: Table Namespace Manager not fully initialized, try again later at org.apache.hadoop.hbase.master.HMaster.checkNamespaceManagerReady(HMaster.java:2693) at org.apache.hadoop.hbase.master.HMaster.ensureNamespaceExists(HMaster.java:2915) at org.apache.hadoop.hbase.master.HMaster.createTable(HMaster.java:1686) Which indicates the AMS HBase master might have some issue. Can you please let us know when was "distributed Mode" AMS running fine last time? Or immediately after enabling AMS distributed mode it is not starting? Is it kerberos enabled environment? Can you please check the permission on the HDFS dir? (to verify if the ownership of this HDFS dir is setup correctly as "ams:hdfs") # su - hdfs -c 'hdfs dfs -ls /user/ams' # su - hdfs -c 'hdfs dfs -ls /user/ams/hbase' If you still face any issue then may be you can try to change the Zookeeper ZNode for AMS and then try restarting AMS freshly. To change the "Zookeeper Znode Parent" property of AMS please try this. Ambari UI --> Amabri Metrics --> Configs --> "Advanced ams-hbase-site" --> "ZooKeeper Znode Parent" then change the value of the znode to something slightly different like "/ams-hbase-unsecure" to "/ams-hbase-unsecure1" ...etc and restart AMS and let us know if you see any error?   ... View more

@mike_bronson7    As the Metrics Service operation mode is already selected to "distributed" hence Ambari will make AMS aware that it needs to find that hbase.rootdir on HDFS.  Following should be fine.   hbase.rootdir=/user/ams/hbase     ... View more

@sharathkumar13    You can use the Cloudera Manager API call something like following to start/stop Kafka Service (or any desired service)   In order to stop "kafka" service:   # curl -iLv -u admin:admin -X POST --header 'Accept: application/json' 'ttp://dc-1.example.com:7180/api/v40/clusters/TestCluster/services/kafka/commands/stop'   In order to start "kafka" Service   # curl -iLv -u admin:admin -X POST --header 'Accept: application/json' 'ttp://dc-1.example.com:7180/api/v40/clusters/TestCluster/services/kafka/commands/start'     . Please also take a look at the Cloudera Manager Swagger APIs.  Cloudera Manager(CM) 6.0 introduces new Python API client cm_client based on Swagger. This new API client supports all CM API versions. https://cloudera.github.io/cm_api/docs/python-client-swagger/ .   Here please replace the CM credentials and cluster name in CM host port in the above mentioned API calls. ... View more