gpload update_columns

`gpload` is a utility in Greenplum Database that is used to load data into database tables in parallel. The `update_columns` parameter in `gpload` specifies the columns that are to be updated instead of inserted when loading data into an existing table. This parameter is used when performing an `UPSERT` operation, where a row is updated if it already exists in the table, and inserted if it doesn't.

The syntax for using `update_columns` in `gpload` is as follows:

UPDATE_COLUMNS ( column_name [, ...] )

Here, `column_name` refers to the name of the column that is to be updated. Multiple columns can be specified by separating them with commas.

For example, the following `gpload` configuration file loads data into a table called `my_table`, and updates the `value` column if the primary key `id` already exists in the table:

LOAD
  TABLE my_table
  (
    id INTEGER,
    value TEXT
  )
  SOURCE (
    FILE='data.csv'
    FORMAT='csv'
    DELIMITER=','
    HEADER='true'
  )
  MAPPING (
    id = 1,
    value = 2
  )
  ERROR_LIMIT = 100
  UPDATE_COLUMNS ( value )
  ;

In this example, the `UPDATE_COLUMNS ( value )` parameter specifies that the `value` column should be updated if a row with the same `id` already exists in the `my_table` table.