Performing Operations on a Hudi Table Using Spark SQL

Scenarios

This section describes how to use Hudi with Spark SQLs.

Prerequisites

You have created a user and added the user to user groups hadoop (primary group) and hive on FusionInsight Manager.

Procedure

1. Download and install the Hudi client. For details, see Installing a Client.
   

   Currently, Hudi is integrated in Spark. You only need to download the Spark client on FusionInsight Manager. For example, the client installation directory is /opt/client.

2. Log in to the node where the client is installed as user root and run the following command:

   cd /opt/client

3. Run the following commands to load environment variables:

   source bigdata_env

   source Hudi/component_env

   kinit Created user

   

   • You need to change the password of the created user, and then run the kinit command to log in to the system again.
   • In normal mode (Kerberos authentication disabled), you do not need to run the kinit command.
   • If multiple services are installed, run the component_env command of the source Spark and then the component_env command of the source Hudi after you run the source bigdata_env command.

4. Start Spark SQL.
   • Create a Hudi table.

     create table if not exists hudi_table2 (id int,name string,price double) using hudi options (type = 'cow',primaryKey = 'id',preCombineField = 'price');

   • Insert data.

     insert into hudi_table2 select 1,1,1;

     insert into hudi_table2 select 2,1,1;

   • Update data.

     update hudi_table2 set name=3 where id=1;

   • Delete data.

     delete from hudi_table2 where id=2;

   • Query data.

     select * from hudi_table2;