RPC service
Main RPC port used by client to communicate with HDFS using a binary protocol. The port is embedded in the URI, eg hdfs://nn1.domain.com:8020/.
Secure data transfert
The datanode server address and port for data transfer. The value depends on the usage of SASL to authenticate data transfer protocol instead of running the DataNode as root, learn more about securing the DataNode.
Port: 9866 (SASL based IPC, non-privileged port) or 1004 (privileged port)
HTTPS service
The datanode secure HTTP server address and port. It is used to access the status, logs, etc, and file data operations when using WebHDFS or HttpFS. The NameNode UI redirects the user to the DataNode server when browsing files.
RPC server
The address of the applications manager interface in the RM. It is used to submit jobs. In YARN non HA configuration yarn.resourcemanager.address uses port 8050. In YARN HA configuration, yarn.resourcemanager.address is redundant and instead yarn.resourcemanager.address.{id} is resolved and uses port 8032.
MapReduce ApplicationMaster
Ephemeral HTTPS ports are opened by each ApplicationMaster. The tdp-collection default port range is unrestricted but is parameterized in the tdp_var_defaults/hadoop/hadoop.yml inventory file under the yarn.app.mapreduce.am.job.client.port-range property. Ports within this range can be accessible from outside the cluster if the permitted by the network firewall. Note that this only restricts the port range used for mapreduce jobs. In the case of Spark, refer to the Spark Driver port documentation, such as the spark.driver.port property.
RPC server
This address for the timeline server to start the RPC server. It addresses the storage and retrieval of application’s current and historic information in a generic fashion.
Default port that the ShuffleHandler will run on. ShuffleHandler is a service run at the NodeManager to facilitate transfers of intermediate Map outputs to requesting Reducers.
Leader server
Peers use the former port to connect to other peers, for example, to agree upon the order of updates. More specifically, a ZooKeeper server uses this port to connect followers to the leader.
Driver port
If the spark.driver.port fails, it is incremented by 1 and retried up to spark.port.maxRetries times.
spark.blockManager.port must be larger than spark.driver.port + sparkspark.sport.maxRetries.
These parameters are identical for Spark2 and Spark3.