GBase 8a, the China‑domestically developed MPP database from GBASE, supports loading data directly from Hadoop HDFS, including clusters secured with Kerberos authentication. This guide covers the configuration parameters, protocol and port selection, and a working load example for your gbase database.
Configuration Parameters
All HDFS parameters must be placed under the [gbased] section of the gcluster and gnode configuration files. They can also be set temporarily via SET for testing before making them permanent.
| Parameter | Description |
|---|---|
gbase_hdfs_protocol |
Protocol type: HTTP, HTTPS, or RPC. |
gbase_hdfs_port |
Port number matching the protocol; differs between Hadoop 2.x and 3.x. |
gbase_hdfs_auth_mode |
Authentication mode: simple (basic) or kerberos. |
gbase_hdfs_namenodes |
Comma‑separated list of NameNode addresses; supports multiple HA groups separated by `\ |
{% raw %}gbase_hdfs_keytab
|
Path to the Kerberos keytab file. |
gbase_hdfs_principal |
Kerberos principal in the format username/hostname@REALM.COM. |
Protocol and Port Reference
| Protocol | Hadoop 2.x Port | Hadoop 3.x Port |
|---|---|---|
| HTTP | 50070 | 9870 |
| HTTPS | 50470 | 9871 |
| RPC | 9000 / 8020 | 8020 |
NameNode Examples
gbase_hdfs_namenodes='192.168.1.1,192.168.1.2'
gbase_hdfs_namenodes='hdfs_nn1:8020,hdfs_nn2:8020'
gbase_hdfs_namenodes='192.168.1.1:50170,192.168.1.2|192.168.2.1,192.168.2.2'
Hosts Configuration
Every node in the cluster must have the Hadoop NameNode and DataNode hostnames and IPs correctly defined in /etc/hosts.
Loading Data from a Kerberised HDFS
Once the parameters are set, load a file with the hdp:// scheme:
gbase> LOAD DATA INFILE 'hdp://hdfs:hdfs2@hadoop141/*.txt'
INTO TABLE t2 FIELDS TERMINATED BY ',';
Query OK, 2 rows affected (Elapsed: 00:00:01.02)
Task 4107 finished, Loaded 2 records, Skipped 0 records
Troubleshooting
If you encounter 401 Authentication required when using HTTP/HTTPS, switch the protocol to RPC. RPC often handles Kerberos in secured Hadoop environments more smoothly than the HTTP‑based alternatives.
With these settings, your gbase database can efficiently pull data from a protected HDFS, keeping your ETL pipelines secure and high‑throughput.

Top comments (0)