PostgreSQL

Syntax

CREATE JOB <job_name>
    [{ job_options }]
    AS COPY FROM POSTGRES <connection_identifier>
         [{ source_options }]
    INTO <table_identifier>
[ WITH EXPECTATION <exp_name> EXPECT <sql_predicate> ON VIOLATION { DROP | WARN } ];   

Jump to

• Job options

• Source options

Job options

[ COLUMN_TRANSFORMATIONS = (<column> = <expression>, ...) ]
[ COMMENT = '<comment>' ]
[ COMPUTE_CLUSTER = <cluster_identifier> ]                     
[ END_AT = { NOW | <timestamp> } ]  
[ EXCLUDE_COLUMNS = ( <col>, ...) ]
[ HEARTBEAT_TABLE = '<heartbeat_name>' ]
[ PARSE_JSON_COLUMNS = { TRUE | FALSE } ]
[ PUBLICATION_NAME = ('regexFilter1', 'regexFilter2') ] 
[ SKIP_SNAPSHOTS = { TRUE | FALSE } ]  
[ SNAPSHOT_PARALLELISM = <integer> ]       

Jump to

PostgreSQL job options:

• HEARTBEAT_TABLE

• PARSE_JSON_COLUMNS

• PUBLICATION_NAME

• SKIP_SNAPSHOTS

• SNAPSHOT_PARALLELISM

General job options:

• COLUMN_TRANSFORMATIONS

• COMMENT

• COMPUTE_CLUSTER

• END_AT

• EXCLUDE_COLUMNS

See also:

• WITH EXPECTATION

HEARTBEAT_TABLE

Type: string

The name of the heartbeat table to use as described in Setting up a Heartbeat Table.

(Optional) If it is not set, no heartbeat table is used. Using a heartbeat table is recommended to avoid the replication slot growing indefinitely when no CDC events are captured for the subscribed tables.

PARSE_JSON_COLUMNS

Type: Boolean

Default: false

If enabled, Upsolver will parse JSON columns into a struct matching the JSON value.

PUBLICATION_NAME

Type: text

Adds a new publication to the current database. The publication name must be distinct from the name of any existing publication in the current database. DDL will be filtered.

SKIP_SNAPSHOTS

Type: Boolean

Default: false

(Optional) By default, snapshots are enabled for new tables. This means that Upsolver will take a full snapshot of the table(s) and ingest it into the staging table before it continues to listen for change events. When True, Upsolver will not take an initial snapshot and only process change events starting from the time the ingestion job is created.

In the majority of cases, when you connect to your source tables, you want to take a full snapshot and ingest it as the baseline of your table. This creates a full copy of the source table in your data lake before you begin to stream the most recent change events. If you skip taking a snapshot, you will not have the historical data in the target table, only the newly added or changed rows.

Skipping a snapshot is useful in scenarios where your primary database instance crashed or became unreachable, failing over to the secondary. In this case, you will need to re-establish the CDC connection but would not want to take a full snapshot because you already have all of the history in your table. In this case, you would want to restart processing from the moment you left off when the connection to the primary database went down.

SNAPSHOT_PARALLELISM

Type: int

Default: 1

(Optional) Configures how many snapshots are performed concurrently. The more snapshots performed concurrently, the quicker the tables are streaming. However, doing more snapshots in parallel increases the load on the source database.

Source Options

[ TABLE_INCLUDE_LIST = ('regexFilter1', 'regexFilter2') ]
[ COLUMN_EXCLUDE_LIST = ('regexFilter1', 'regexFilter2') ]          

TABLE_INCLUDE_LIST — editable

Type: text

Default: ''

(Optional) Comma-separated list of regular expressions that match fully-qualified table identifiers of tables whose changes you want to capture. Tables not included in this list will not be loaded. If the list is left empty all tables will be loaded. This maps to the Debezium table.include.list property.

By default, the connector captures changes in every non-system table in all databases. To match the name of a table, Upsolver applies the regular expression that you specify as an anchored regular expression. That is, the specified expression is matched against the entire name string of the table. It does not match substrings that might be present in a table name.

Each RegEx pattern matches against the full string databaseName.tableName, for example:

RegEx Pattern	Results
db_name.*	Select all tables from the db_name database.
db_name.users, db_name.items	Select the users and items tables from the db_name database.
db1.items_.*	Select all tables from db1 that start with items_.

COLUMN_EXCLUDE_LIST — editable

Type: array[string]

Default: ''

(Optional) Comma-separated list of regular expressions that match the fully-qualified names of columns to exclude from change event record values. This maps to the Debezium column.exclude.list property.

By default, the connector matches all columns of the tables listed in TABLE_INCLUDE_LIST. To match the name of a column, Upsolver applies the regular expression that you specify as an anchored regular expression. That is, the specified expression is matched against the entire name string of the column; it does not match substrings that might be present in a column name.

Each RegEx pattern matches against the full string databaseName.tableName.columnName, for example:

RegEx Pattern	Results
db.users.address_.*	Select all columns starting with address_ from the users table in the db database.
db.*.(.*_pii)	Select all columns ending in _pii across all tables in the db database.

Examples

Ingest data into the data lake

The following example creates a job to ingest data from PostgreSQL into a table in the data lake. The PUBLICATION_NAME option specifies that a new publication named sample is added.

CREATE JOB load_orders_raw_data_from_postgres
      PUBLICATION_NAME = 'sample'
AS COPY FROM POSTGRES upsolver_postgres_samples 
INTO default_glue_catalog.upsolver_samples.orders_raw_data;