5.4. Configure ports

HDP uses multiple ports for communication with clients and between service components. To enable this communication, you will need to either open all ports or the specific ports that HDP uses.

To open specific ports only, you can set the access rules in Windows.

For example, the following command will open up port 80 in the active Windows Firewall:

netsh advfirewall firewall add rule name=AllowRPCCommunication dir=in action=allow protocol=TCP localport=135

For example, the following command will open up ports 49152-65535 in the active Windows Firewall:

netsh advfirewall firewall add rule name=AllowRPCCommunication dir=in action=allow protocol=TCP localport=49152-65535

The tables below specify which ports must be opened for which ecosystem components to communicate with each other.

Make sure that appropriate ports are opened before you install HDP.

HDFS Ports: The following table lists the default ports used by the various HDFS services.

Table 1.1. HDFS Ports
Service Servers Default Ports Used Protocol Description Need End User Access? Configuration Parameters

NameNode WebUI

Master Nodes (NameNode and any back-up NameNodes) 50070 http Web UI to look at current status of HDFS, explore file system Yes (Typically admins, Dev/Support teams) dfs.http.address
50470 https Secure http service dfs.https.address

NameNode metadata service

8020/9000 IPC

File system metadata operations

Yes (All clients who directly need to interact with the HDFS) Embedded in URI specified by fs.default.name

DataNode

All Slave Nodes

50075

http

DataNode WebUI to access the status, logs etc.

Yes (Typically admins, Dev/Support teams) dfs.datanode.http.address

50475

https

Secure http service

dfs.datanode.https.address

50010

Data transfer

dfs.datanode.address

50020

IPC

Metadata operations

No dfs.datanode.ipc.address
Secondary NameNode Secondary NameNode and any backup Secondanry NameNode

50090

http

Checkpoint for NameNode metadata

No dfs.secondary.http.address

MapReduce Ports: The following table lists the default ports used by the various MapReduce services.

Table 1.2. MapReduce Ports
Service Servers Default Ports Used Protocol Description Need End User Access? Configuration Parameters

JobTracker  WebUI

Master Nodes (JobTracker Node and any back-up Job­Tracker node ) 50030 http Web UI for JobTracker Yes mapred.job.tracker.http.address

JobTracker

Master Nodes (JobTracker Node) 8021 IPC

For job submissions

Yes (All clients who need to submit the MapReduce jobs  including Hive, Hive server, Pig) Embedded in URI specified by mapred.job.tracker

Task­Tracker Web UI and Shuffle

All Slave Nodes

50060

http DataNode Web UI to access status, logs, etc. Yes (Typically admins, Dev/Support teams) mapred.task.tracker.http.address
History Server WebUI 51111 http Web UI for Job History Yes mapreduce.history.server.http.address

Hive Ports: The following table lists the default ports used by the Hive services.

[Note]Note

These services are not used in a standard HDP installation.

Table 1.3. Hive Ports
Service Servers Default Ports Used Protocol Description Need End User Access? Configuration Parameters

HiveServer2

HiveServer2 machine (Usually a utility machine) 10001 thrift Service for programmatically (Thrift/JDBC) connecting to Hive Yes ENV Variable HIVE_PORT

Hive Server

Hive Server machine (Usually a utility machine) 10000 thrift Service for programmatically (Thrift/JDBC) connecting to Hive Yes (Clients who need to connect to Hive either programatically or through UI SQL tools that use JDBC) ENV Variable HIVE_PORT

Hive Metastore

9083 thrift Service for programmatically (Thrift/JDBC) connecting to Hive metadata Yes (Clients that run Hive, Pig and potentially M/R jobs that use HCatalog) hive.metastore.uris

Templeton Port: The following table lists the default port used by the Templeton service.

Table 1.4. Templeton Port
Service Servers Default Ports Used Protocol Description Need End User Access? Configuration Parameters

Templeton Server

Any utility machine 50111 http Web API on top of HCatalog and other Hadoop services Yes templeton.port