About PXF Server Configuration

A newer version of this documentation is available. Use the version menu above to view the most up-to-date release of the Greenplum 5.x documentation.

This topic provides an overview of PXF server configuration. To configure a server, refer to the topic specific to the connector that you want to configure.

You read from or write data to an external data store via a PXF connector. To access an external data store, you must provide the server location. You may also be required to provide client access credentials and other external data store-specific properties. PXF simplifies configuring access to external data stores by:

  • Supporting file-based configuration
  • Providing connector-specific template configuration files

A PXF Server definition is simply a named configuration that provides access to a specific external data store as a specific user. A PXF server name is the name of a directory residing in $PXF_CONF/servers/. The information that you provide in a server configuration is connector-specific. PXF provides a server template file for each connector; this template identifies the minimum set of properties that you must configure to use the connector.

You configure a server definition for each external data store that you will permit Greenplum Database users to access. For example, if you require access to two Hadoop clusters, you will create a PXF Hadoop server configuration for each cluster. If you require access to an Oracle and a MySQL database, you will create a PXF JDBC server configuration for each database/user combination.

After you configure a PXF server, you publish the server name to Greenplum Database users as appropriate. A user only needs to provide the server name when they create an external table that accesses the external data store. PXF obtains the external data source location and access credentials from the server configuration directory identified by the server name.

Server Configuration Procedure

When you configure a PXF connector, you add a named PXF server configuration for the connector. Among the tasks that you perform, you may:

  1. Determine if you are configuring the default PXF server, or choose a new name for the server configuration.
  2. Create the directory $PXF_CONF/servers/<server_name>.
  3. Copy template or other configuration files to the new server directory.
  4. Fill in appropriate values for the properties in the template file.
  5. Add additional properties and values if required for your environment.
  6. Add other configuration information supported by the connector.
  7. Synchronize the server configuration to the Greenplum Database cluster.
  8. Publish the PXF server names to your Greenplum Database end users as appropriate.

About The Default Server

PXF defines a special server named default. When you initialize PXF, it automatically creates a $PXF_CONF/servers/default/ directory. This directory, initially empty, identifies the default PXF server configuration. You can configure and assign the default PXF server to any external data source. For example, you may choose to assign the PXF default server to a Hadoop cluster, or to a MySQL database that you frequently access as a specific user.

Note: You must configure a Hadoop server as the PXF default server when your Hadoop cluster utilizes Kerberos authentication.

PXF automatically uses the default server configuration if you omit the SERVER=<server_name> setting in the CREATE EXTERNAL TABLE command LOCATION clause.

Server Template Files

PXF provides a template configuration file for each connector. These server template configuration files, located in the $PXF_CONF/templates/ directory, identify the minimum set of properties that you must configure to use the connector.

gpadmin@gpmaster$ ls $PXF_CONF/templates
adl-site.xml   hbase-site.xml  jdbc-site.xml    s3-site.xml
core-site.xml  hdfs-site.xml   mapred-site.xml  wasbs-site.xml
gs-site.xml    hive-site.xml   minio-site.xml   yarn-site.xml

For example, the contents of the s3-site.xml template file follow:

<?xml version="1.0" encoding="UTF-8"?>
You specify credentials to PXF in clear text in configuration files.

The template files for the Hadoop connectors are not intended to be modified and used for configuration, as they only provide an example of the information needed. Instead of modifying the Hadoop templates, you will copy several Hadoop *-site.xml files from the Hadoop cluster to your PXF Hadoop server configuration.

Using a Server Configuration

To access an external data store, the Greenplum Database user specifies the server name in the CREATE EXTERNAL TABLE command LOCATION clause SERVER=<server_name> option. The <server_name> that the user provides identifies the server configuration directory from which PXF obtains the configuration and credentials to access the external data store.

For example, the following command accesses an S3 object store using the server configuration defined in $PXF_CONF/servers/s3srvcfg:

CREATE EXTERNAL TABLE pxf_ext_tbl(name text, orders int)
  LOCATION ('pxf://BUCKET/dir/file.txt?PROFILE=s3:text&SERVER=s3srvcfg')
FORMAT 'TEXT' (delimiter=E',');

PXF automatically uses the default server configuration when no SERVER=<server_name> setting is provided.

For example, if the default server configuration identifies a Hadoop cluster, the following example command references the HDFS file located at /path/to/file.txt:

CREATE EXTERNAL TABLE pxf_ext_hdfs(location text, miles int)
  LOCATION ('pxf://path/to/file.txt?PROFILE=hdfs:text')
FORMAT 'TEXT' (delimiter=E',');
A Greenplum Database user who queries or writes to an external table accesses the external data store with the credentials configured for the server associated with the external table.

Adding or Updating Server Configurations

If you add new, or update existing, PXF server configurations on the Greenplum Database master host, you must re-sync the PXF configuration to the Greenplum Database cluster:

gpadmin@gpmaster$ $GPHOME/pxf/bin/pxf cluster sync

Overriding the Server Configuration

Some PXF connectors (S3, JDBC) allow you to override a server configuration by directly specifying certain properties via custom options in the CREATE EXTERNAL TABLE command LOCATION clause. Refer to Overriding the S3 Server Configuration and Overriding the JDBC Server Configuration for additional information.

When you override a server configuration, the properties and their values are visible as part of the external table defininition. Do not use this method to pass credentials in a production environment.