In ambari webui there is an option to start all the services with one click, Some times that start all option will not work properly due to some technical challenges then what is the correct order to start hadoop services in ambari manually is the biggest confusion. To avoid this confusion i am going to explain you the correct order to start hadoop services.
Order to Start Hadoop Services in Ambari Manually
Start the Hadoop services in the following order by using Ambari Webui.
- Hive Metastore
Better to start the services with ranger first,the reason is ranger doesn’t depend on Namenode,zookeeper and other services until ranger don’t have high availability.
Knox service also doesn’t depend on Namenode,zookeeper and other services to start it’s components.
Before to start any high availability services zookeeper should up and run, so it’s compulsory to start zookeeper before start HDFS,YARN,HIVE,HBASE etc services.
Once zookeeper up,we can start hdfs service,so no probelm to journal nodes,zkfcontroller services and high availability.
If Zookeeper and HDFS both services are up and running fine,then no problem to start Yarn service. No impact to Resource manager high availability.
Hbase high availability also depends on Zookeeper service.
Before start hive server2 service hive metastore should up and running.
If hive metastore is up and running then no issues to start hiveserver2 services.
If Both hive metastore and hiveserver2 services are running fine,then no problem to start webhcat service.
Oozie depends on critical services like hdfs,yarn,hive,hbase etc. So,If that services are up and running we are good to start Oozie service.
By using Hue webui hadoop users can access hdfs,hive,hbase,oozie,mapreduce services
Strom,Kafka,Atlas services we can run it by order.so, this is the Correct Order to Start Hadoop Services in Ambari Manually.