Create an OpenSearch sink connector
Use an OpenSearch sink connector to make the changes captured from your source database searchable in OpenSearch. The connector reads the Kafka topics you choose and writes each topic to an OpenSearch index.
Prerequisites
- A CDC Service whose status is Healthy. See Create a CDC Service.
- Topics with change data, written by a source connector such as a PostgreSQL source connector.
- The connection string, username, and password of your OpenSearch database.
Create the connector
-
Select Data platform → Workspace management, then select your workspace.
-
On the Overview tab, under Services, select the CDC card.
-
Select the Connectors tab, then select Create Connector.

-
On the Connector Information screen, enter a Name, select sink in Type, and select OpenSearch in Database.

-
Select Next Step. On the Properties screen, enter the connection details, then select Test connection. Next Step becomes available after the test succeeds. If the test fails with "Connection failed due to network failure or invalid information.", check the connection string, the username, and the password, then test again.


-
On the same screen, choose a Converter class and set the four converter fields, then select the topics to read in Topics.


-
Select Next Step. On the Additional properties screen, set the fields in the Task, Map table, and Error Tolerance sections.



-
Select Next Step and check the settings on the Review screen.

-
Select Create.
The portal returns to the Essential Information tab of the CDC Service. On the Connectors tab, the new connector shows the status Running, with every Tasks counter at 0 until its task starts, which can take up to about 2.5 minutes.
Connector settings
Connector information
| Field | Required | Description |
|---|---|---|
| Name | Yes | The connector name. Use lowercase letters, digits, and hyphens (-) only. Upper-case letters, _, ., and spaces are rejected with "Name has invalid characters". |
| Type | Yes | Select sink. |
| Database | Yes | Select OpenSearch. |
Properties
In the Database Information section, choose From FPT Database Engine or Manual configuration. Manual configuration is selected by default. With From FPT Database Engine, select your instance in Database Name. The portal fills Connection string as http://HOST:9200, which you cannot edit. See Connection mode.
| Field | Required | Description |
|---|---|---|
| Database Name | Yes | Shown with From FPT Database Engine only. The OpenSearch instance in FPT Database Engine. |
| Connection string | Yes | The connection URI or host name of your OpenSearch database. |
| Username | Yes | The username the connector uses to connect to OpenSearch. |
| Password | Yes | The password for that user. |
| Converter class | Yes | The converter class for this connector. The portal offers Opensearch. The four converter fields stay greyed out until you choose a class. See Converter. |
| Converter key, Converter key schema enable, Converter value, Converter value schema enable | Yes | How the connector reads the keys and values of the messages, and whether they carry a schema. Converter key offers org.apache.kafka.connect.storage.StringConverter and org.apache.kafka.connect.json.JsonConverter, and Converter value only org.apache.kafka.connect.json.JsonConverter. Both schema enable fields offer true and false. |
| Topics (under Kafka topic) | Yes | The topics to read. Select them from the list, which shows the topics in the Kafka cluster of your CDC Service. See Topics. |
Additional properties
| Field | Required | Description |
|---|---|---|
| Tasks max | Yes | The number of tasks the connector runs at the same time when the topics have more than one partition. Above 1, records can be consumed out of order, so give related records the same key to keep them in one partition. Default 1. See Tasks max. |
| Data stream enable | No | Turn on to write to a data stream instead of a regular index. Data streams suit time-series and log data: they only append documents, and each record needs a @timestamp field. Default Disable. |
| Topic 1, Create new table, Table 1 (under Map table) | Yes | The OpenSearch index that receives the records of each topic. Despite its label, Table 1 takes an index name. Create new table is off by default, and Table 1 lists the existing indices, including system indices whose names start with .. Do not select a system index. Turn on Create new table to type a new index name instead. Table 1 is then empty. See Map table. |
| Mode (under Error Tolerance) | Yes | What the connector does with a message that it cannot read or convert. None stops the connector with status Failed. All skips the message and sends it to the topic you enter for failed records. The field has no default. See Error tolerance. |