StarRocks

StarRocks sink connector

Support These Engines

Spark
Flink
SeaTunnel Zeta

Key Features

• exactly-once
• cdc

Description

Used to send data to StarRocks. Both support streaming and batch mode. The internal implementation of StarRocks sink connector is cached and imported by stream load in batches.

Sink Options

Name	Type	Required	Default	Description
nodeUrls	list	yes	-	StarRocks cluster address, the format is ["fe_ip:fe_http_port", ...]
base-url	string	yes	-	The JDBC URL like jdbc:mysql://localhost:9030/ or jdbc:mysql://localhost:9030 or jdbc:mysql://localhost:9030/db
username	string	yes	-	StarRocks user username
password	string	yes	-	StarRocks user password
database	string	yes	-	The name of StarRocks database
table	string	no	-	The name of StarRocks table, If not set, the table name will be the name of the upstream table
labelPrefix	string	no	-	The prefix of StarRocks stream load label
batch_max_rows	long	no	1024	For batch writing, when the number of buffers reaches the number of batch_max_rows or the byte size of batch_max_bytes or the time reaches checkpoint.interval, the data will be flushed into the StarRocks
batch_max_bytes	int	no	5 1024 1024	For batch writing, when the number of buffers reaches the number of batch_max_rows or the byte size of batch_max_bytes or the time reaches checkpoint.interval, the data will be flushed into the StarRocks
max_retries	int	no	-	The number of retries to flush failed
retry_backoff_multiplier_ms	int	no	-	Using as a multiplier for generating the next delay for backoff
max_retry_backoff_ms	int	no	-	The amount of time to wait before attempting to retry a request to StarRocks
enable_upsert_delete	boolean	no	false	Whether to enable upsert/delete, only supports PrimaryKey model.
save_mode_create_template	string	no	see below	see below
starrocks.config	map	no	-	The parameter of the stream load data_desc
http_socket_timeout_ms	int	no	180000	Set http socket timeout, default is 3 minutes.
schema_save_mode	Enum	no	CREATE_SCHEMA_WHEN_NOT_EXIST	Before the synchronous task is turned on, different treatment schemes are selected for the existing surface structure of the target side.
data_save_mode	Enum	no	APPEND_DATA	Before the synchronous task is turned on, different processing schemes are selected for data existing data on the target side.
custom_sql	String	no	-	When data_save_mode selects CUSTOM_PROCESSING, you should fill in the CUSTOM_SQL parameter. This parameter usually fills in a SQL that can be executed. SQL will be executed before synchronization tasks.

save_mode_create_template

We use templates to automatically create starrocks tables, which will create corresponding table creation statements based on the type of upstream data and schema type, and the default template can be modified according to the situation. Only work on multi-table mode at now.

table [string]

Use database and this table-name auto-generate sql and receive upstream input datas write to database.

This option is mutually exclusive with query and has a higher priority.

The table parameter can fill in the name of an unwilling table, which will eventually be used as the table name of the creation table, and supports variables (${table_name}, ${schema_name}). Replacement rules: ${schema_name} will replace the SCHEMA name passed to the target side, and ${table_name} will replace the name of the table passed to the table at the target side.

for example:

1. test${schema_name}${table_name}_test
2. sink_sinktable
3. ss_${table_name}

schema_save_mode[Enum]

Before the synchronous task is turned on, different treatment schemes are selected for the existing surface structure of the target side.
Option introduction：
RECREATE_SCHEMA ：Will create when the table does not exist, delete and rebuild when the table is saved
CREATE_SCHEMA_WHEN_NOT_EXIST ：Will Created when the table does not exist, skipped when the table is saved
ERROR_WHEN_SCHEMA_NOT_EXIST ：Error will be reported when the table does not exist

data_save_mode[Enum]

Before the synchronous task is turned on, different processing schemes are selected for data existing data on the target side.
Option introduction：
DROP_DATA： Preserve database structure and delete data
APPEND_DATA：Preserve database structure, preserve data
CUSTOM_PROCESSING：User defined processing
ERROR_WHEN_DATA_EXISTS：When there is data, an error is reported

custom_sql[String]

When data_save_mode selects CUSTOM_PROCESSING, you should fill in the CUSTOM_SQL parameter. This parameter usually fills in a SQL that can be executed. SQL will be executed before synchronization tasks.

CREATE TABLE IF NOT EXISTS `${database}`.`${table_name}`
(
    ${rowtype_fields}
) ENGINE = OLAP DISTRIBUTED BY HASH (${rowtype_primary_key})
    PROPERTIES
(
    "replication_num" = "1"
);

If a custom field is filled in the template, such as adding an id field

CREATE TABLE IF NOT EXISTS `${database}`.`${table_name}`
(   
    id,
    ${rowtype_fields}
) ENGINE = OLAP DISTRIBUTED BY HASH (${rowtype_primary_key})
    PROPERTIES
(
    "replication_num" = "1"
);

The connector will automatically obtain the corresponding type from the upstream to complete the filling, and remove the id field from rowtype_fields. This method can be used to customize the modification of field types and attributes.

You can use the following placeholders

• database: Used to get the database in the upstream schema
• table_name: Used to get the table name in the upstream schema
• rowtype_fields: Used to get all the fields in the upstream schema, we will automatically map to the field description of StarRocks
• rowtype_primary_key: Used to get the primary key in the upstream schema (maybe a list)
• rowtype_unique_key: Used to get the unique key in the upstream schema (maybe a list)

Data Type Mapping

StarRocks Data type	SeaTunnel Data type
BOOLEAN	BOOLEAN
TINYINT	TINYINT
SMALLINT	SMALLINT
INT	INT
BIGINT	BIGINT
FLOAT	FLOAT
DOUBLE	DOUBLE
DECIMAL	DECIMAL
DATE	STRING
TIME	STRING
DATETIME	STRING
STRING	STRING
ARRAY	STRING
MAP	STRING
BYTES	STRING

Supported import data formats

The supported formats include CSV and JSON

Task Example

Simple:

The following example describes writing multiple data types to StarRocks, and users need to create corresponding tables downstream

env {
  parallelism = 1
  job.mode = "BATCH"
  checkpoint.interval = 10000
}

source {
  FakeSource {
    row.num = 10
    map.size = 10
    array.size = 10
    bytes.length = 10
    string.length = 10
    schema = {
      fields {
        c_map = "map<string, array<int>>"
        c_array = "array<int>"
        c_string = string
        c_boolean = boolean
        c_tinyint = tinyint
        c_smallint = smallint
        c_int = int
        c_bigint = bigint
        c_float = float
        c_double = double
        c_decimal = "decimal(16, 1)"
        c_null = "null"
        c_bytes = bytes
        c_date = date
        c_timestamp = timestamp
      }
    }
    }
}

sink {
  StarRocks {
    nodeUrls = ["e2e_starRocksdb:8030"]
    username = root
    password = ""
    database = "test"
    table = "e2e_table_sink"
    batch_max_rows = 10
    starrocks.config = {
      format = "JSON"
      strip_outer_array = true
    }
  }
}

Support write cdc changelog event(INSERT/UPDATE/DELETE)

sink {
  StarRocks {
    nodeUrls = ["e2e_starRocksdb:8030"]
    username = root
    password = ""
    database = "test"
    table = "e2e_table_sink"
    ...
    
    // Support upsert/delete event synchronization (enable_upsert_delete=true), only supports PrimaryKey model.
    enable_upsert_delete = true
  }
}

Use JSON format to import data

sink {
  StarRocks {
    nodeUrls = ["e2e_starRocksdb:8030"]
    username = root
    password = ""
    database = "test"
    table = "e2e_table_sink"
    batch_max_rows = 10
    starrocks.config = {
      format = "JSON"
      strip_outer_array = true
    }
  }
}

Use CSV format to import data

sink {
  StarRocks {
    nodeUrls = ["e2e_starRocksdb:8030"]
    username = root
    password = ""
    database = "test"
    table = "e2e_table_sink"
    batch_max_rows = 10
    starrocks.config = {
      format = "CSV"
      column_separator = "\\x01"
      row_delimiter = "\\x02"
    }
  }
}

Use save_mode function

sink {
  StarRocks {
    nodeUrls = ["e2e_starRocksdb:8030"]
    username = root
    password = ""
    database = "test"
    table = "test_${schema_name}_${table_name}"
    schema_save_mode = "CREATE_SCHEMA_WHEN_NOT_EXIST"
    data_save_mode="APPEND_DATA"
    batch_max_rows = 10
    starrocks.config = {
      format = "CSV"
      column_separator = "\\x01"
      row_delimiter = "\\x02"
    }
  }
}

Changelog

next version

• Add StarRocks Sink Connector
• [Improve] Change Connector Custom Config Prefix To Map 3719
• [Feature] Support write cdc changelog event(INSERT/UPDATE/DELETE) 3865