IBM Fluid Query 1.7.2

Registering the data connector functions

To call and use the data connector functions in your queries, you must register the data connector functions in one or more NPS® databases.

Before you begin

After you have created at least one connection to your Hadoop service providers, you can then register functions in a database. Typically, you register the functions only once in each NPS database that is used for SQL select queries with the Hadoop tables.

Procedure

Log in to the NPS active host as the nz user.
Change to the /nz/export/ae/products/fluidquery/ directory.
Run the fqRegister.sh script to register the functions. There are four data connector function definitions that have the same function name but different signatures to support different types of query invocations. By default, the data connector functions use the name FqRead(), but you can register the functions under a unique name for your environment. Sample commands follow.
- To perform a simple registration where you add the FqRead() functions to the database specified by NZ_DATABASE (which is set to a database named maindb) using the connection defined in the default.properties configuration:
```
./fqRegister.sh
Functions and credentials are successfully registered in database "MAINDB".
```
- To register the functions using the name HdpRead() in the database named MyDb using the connection defined in themyConfig.properties connection configuration:
```
./fqRegister.sh --udtf HdpRead --db MyDb --config myConfig
Functions and credentials are successfully registered in database "MYDB".
```

Results

The registration function adds the data connector functions to the specified database. Each registration requires a connection configuration. You must specify a connection configuration file that was created by the fqConfigure.sh script, or the command uses the default connection called default.properties. The functions are owned by the admin user, and for other database users to use the functions in their queries, you must grant those users privileges to execute the functions. For more information about privileges, see Assigning privileges to run the data connector functions.

After registering the functions, your NPS users who have privileges to the database and to execute the functions can include them in their SQL queries. To confirm that the functions were added to the database, you can use the SHOW FUNCTION command to display the functions in your database. A sample command follows. Note that the Arguments column is truncated in the documentation because of the wide length of the field.

MYDB.ADMIN(MYUSER)=> SHOW FUNCTION hdpread;
 SCHEMA |   RESULT   | FUNCTION | BUILTIN |                   ARGUMENTS
--------+------------+----------+---------+---------------------------------------------------
 ADMIN  | TABLE(ANY) | HDPREAD  | f       | (CHARACTER VARYING(ANY), CHARACTER VARYING(ANY))
 ADMIN  | TABLE(ANY) | HDPREAD  | f       | (CHARACTER VARYING(ANY), CHARACTER VARYING(ANY), CH...
 ADMIN  | TABLE(ANY) | HDPREAD  | f       | (CHARACTER VARYING(ANY), CHARACTER VARYING(ANY), CH...
 ADMIN  | TABLE(ANY) | HDPREAD  | f       | (CHARACTER VARYING(ANY), CHARACTER VARYING(ANY), CH...
(4 rows)

If you plan to use the remote mode for your data connector functions, you must register a data connector function with the --remote flag, otherwise the remote service will not start. See Remote mode for more information.

If the command fails and displays an error message, you can try re-running command and adding the --debug option for more troubleshooting information. If you use the --debug option when you register the functions, the software creates log files each time a query runs and calls the functions. The log files can help you to troubleshoot any query problems, but when your queries are operating as expected, make sure that you re-register the functions without the --debug option to stop the log files. For more information about the script and its options, see The fqRegister script.

For more information about the FqRead() function and the four supported input forms, see The FqRead function.

Feedback