recently, Ambari Metrics Collector has been configured with distributed data storage. The changed configuration is as follows:

hbase.cluster.distributed=true
Metrics Service operation mode = distributed
// HDFSams${name}HDFS
hbase.rootdir=hdfs://${name}/user/ams/hbase

starts, it will hang up. Check the / ambari-metrics-collector.log log and find that:
2018-09-03 22 code 00 INFO org.apache.hadoop.hbase.client.ZooKeeperRegistry: ClusterId read in ZooKeeper is null
2018-09-03 22 purse 05code 50762 WARN org.apache.hadoop.yarn.server.applicationhistoryservice.metrics.timeline.query.DefaultPhoenixDataSource: Unable to connect to HBase store using Phoenix.
java.sql.SQLException: org.apache.hadoop.hbase.client.RetriesExhaustedException: Can"t get the locations
.

Open the DEBUG level log and find that during the startup process, the message for a long time to swipe is:
2018-09-26 17 DEBUG org.apache.hadoop.hbase.zookeeper.ZKUtil: hconnection-0x2462cb01 02purl 11127 DEBUG org.apache.hadoop.hbase.zookeeper.ZKUtil: hconnection-0x2462cb01-0x264921b18e6b284, quorum=dfs1a1.ecld.com:2181,dfs1m2.ecld.com:2181,dfs1m1.ecld.com:2181, baseZNode=/ams-hbase-secure Unable to get data of znode / ams-hbase-secure/meta-region-server because node does not exist (not an error)

after checking, it is found that the Phoenix component is not installed, and after installation, the error remains the same.
Log in to ZooKeeper, and find that there is only one / tokenauth node under / ams-hbase-secure , and there is no hbase information.
according to the practice on the Internet, cleaning up tmp, and resetting ZNode, etc., have been tried, but to no avail. Each time under the ams-hbase node configured above the ZK, only one / tokenauth, is missing other information.

ask the gods, ask for advice, thank you!