Skip to main content
Version: 2.5

ALTER ROUTINE LOAD

Description

Alters a Routine Load job that is in the PAUSED state. You can execute PAUSE ROUTINE LOAD to pause a Routine Load job.

After successfully altering a Routine Load job, you can:

Syntax

ALTER ROUTINE LOAD FOR [<db_name>.]<job_name>
[load_properties]
[job_properties]
FROM data_source
[data_source_properties]

Parameters

  • [<db_name>.]<job_name>

    • db_name: Optional. The name of the StarRocks database.
    • job_name: Required. The name of the Routine Load job to be altered.
  • load_properties

    The properties of the source data to be loaded. The syntax is as follows:

    [COLUMNS TERMINATED BY '<column_separator>'],
    [ROWS TERMINATED BY '<row_separator>'],
    [COLUMNS ([<column_name> [, ...] ] [, column_assignment [, ...] ] )],
    [WHERE <expr>],
    [PARTITION ([ <partition_name> [, ...] ])]
    [TEMPORARY PARTITION (<temporary_partition1_name>[,<temporary_partition2_name>,...])]

    For detailed parameter descriptions, see CREATE ROUTINE LOAD.

  • job_properties

    The properties of the load job. The syntax is as follows:

    PROPERTIES ("<key1>" = "<value1>"[, "<key2>" = "<value2>" ...])

    Only the following parameters can be altered:

    • desired_concurrent_number

    • max_error_number

    • max_batch_interval

    • max_batch_rows

    • max_batch_size

    • jsonpaths

    • json_root

    • strip_outer_array

    • strict_mode

    • timezone

    For detailed parameter descriptions, see CREATE ROUTINE LOAD.

  • data_source and data_source_properties

    • data_source

      Required. The source of the data you want to load. Valid value: KAFKA.

    • data_source_properties

      The properties of the data source. Currently, only the following properties can be altered:

      • kafka_partitions and kafka_offsets: Note that StarRocks only supports modifying the offset of Kafka partitions that have already been consumed but does not support adding new Kafka partitions.
      • property.*: Custom parameters for the data source Kafka, such as property.kafka_default_offsets.

Examples

  1. The following example increases the value of the property desired_concurrent_number of the load job to 5 in order to increase the parallelism of load tasks. For details on task parallelism, see how to improve load performance.

    ALTER ROUTINE LOAD FOR example_tbl_ordertest
    PROPERTIES
    (
    "desired_concurrent_number" = "5"
    );
  2. The following example alters the load job's properties and data source information at the same time.

    ALTER ROUTINE LOAD FOR example_tbl_ordertest
    PROPERTIES
    (
    "desired_concurrent_number" = "5"
    )
    FROM KAFKA
    (
    "kafka_partitions" = "0, 1, 2",
    "kafka_offsets" = "100, 200, 100",
    "property.group.id" = "new_group"
    );
  3. The following example alters the filtering condition and the StarRocks partitions that data is loaded into at the same time.

    ALTER ROUTINE LOAD FOR example_tbl_ordertest
    WHERE pay_dt < 2023-06-31
    PARTITION (p202306);