Hudi

Hudi source connector

Description

Used to read data from Hudi. Currently, only supports hudi cow table and Snapshot Query with Batch Mode.

In order to use this connector, You must ensure your spark/flink cluster already integrated hive. The tested hive version is 2.3.9.

Key features

• [x] batch

Currently, only supports hudi cow table and Snapshot Query with Batch Mode

• [ ] stream
• [x] exactly-once
• [ ] schema projection
• [x] parallelism
• [ ] support user-defined split

Options

name	type	required	default value
table.path	string	yes	-
table.type	string	yes	-
conf.files	string	yes	-
use.kerberos	boolean	no	false
kerberos.principal	string	no	-
kerberos.principal.file	string	no	-

table.path [string]

table.path The hdfs root path of hudi table,such as ‘hdfs://nameserivce/data/hudi/hudi_table/’.

table.type [string]

table.type The type of hudi table. Now we only support ‘cow’, ‘mor’ is not support yet.

conf.files [string]

conf.files The environment conf file path list(local path), which used to init hdfs client to read hudi table file. The example is ‘/home/test/hdfs-site.xml;/home/test/core-site.xml;/home/test/yarn-site.xml’.

use.kerberos [boolean]

use.kerberos Whether to enable Kerberos, default is false.

kerberos.principal [string]

kerberos.principal When use kerberos, we should set kerberos princal such as ‘test_user@xxx’.

kerberos.principal.file [string]

kerberos.principal.file When use kerberos, we should set kerberos princal file such as ‘/home/test/test_user.keytab’.

Examples

source {

  Hudi {
    table.path = "hdfs://nameserivce/data/hudi/hudi_table/"
    table.type = "cow"
    conf.files = "/home/test/hdfs-site.xml;/home/test/core-site.xml;/home/test/yarn-site.xml"
    use.kerberos = true
    kerberos.principal = "test_user@xxx"
    kerberos.principal.file = "/home/test/test_user.keytab"
  }

}