Specify HDFS Data in an External Table Definition

The LOCATION clause of the CREATE EXTERNAL TABLE command for HDFS files differs slightly for Hadoop HA (High Availability) clusters, Hadoop clusters without HA, and MapR clusters.

In a Hadoop HA cluster, the LOCATION clause references the logical nameservices id (the dfs.nameservices property in the hdfs-site.xml configuration file). The hdfs-site.xml file with the nameservices configuration must be installed on the Greenplum master and on each segment host.

For example, if dfs.nameservices is set to mycluster the LOCATION clause takes this format:

LOCATION ('gphdfs://mycluster/path/filename.txt')

A cluster without HA specifies the hostname and port of the name node in the LOCATION clause:

LOCATION ('gphdfs://hdfs_host[:port]/path/filename.txt')

If you are using MapR clusters, you specify a specific cluster and the file:

To specify the default cluster, the first entry in the MapR configuration file /opt/mapr/conf/mapr-clusters.conf, specify the location of your table with this syntax:
```
LOCATION ('gphdfs:///file_path')
```
The file_path is the path to the file.
To specify another MapR cluster listed in the configuration file, specify the file with this syntax:
```
LOCATION ('gphdfs:///mapr/cluster_name/file_path')
```
The cluster_name is the name of the cluster specified in the configuration file and file_path is the path to the file.

For information about MapR clusters, see the MapR documentation.

Restrictions for HDFS files are as follows.

You can specify one path for a readable external table with gphdfs. Wildcard characters are allowed. If you specify a directory, the default is all files in the directory.
You can specify only a directory for writable external tables.
Format restrictions are as follows.
- Only the gphdfs_import formatter is allowed for readable external tables with a custom format.
- Only the gphdfs_export formatter is allowed for writable external tables with a custom format.
- You can set compression only for writable external tables. Compression settings are automatic for readable external tables.