Terminate a ksqlDB cluster in ksqlDB for Confluent Platform

If you don’t need your ksqlDB cluster anymore, you can terminate the cluster and clean up the resources using this endpoint. To terminate a ksqlDB cluster, first shut down all of the servers, except one. Then, send the TERMINATE CLUSTER request to the /ksql/terminate endpoint in the last remaining server. When the server receives a TERMINATE CLUSTER request at /ksql/terminate endpoint, it writes a TERMINATE CLUSTER command into the command topic. Note that a TERMINATE CLUSTER request can be sent only via the /ksql/terminate endpoint, and you can’t send it via the CLI. When the server reads the TERMINATE CLUSTER command, it takes the following steps:

  • Sets the ksqlDB engine mode to NOT ACCEPTING NEW STATEMENTS so no new statements are passed to the engine for execution.
  • Terminates all persistent and transient queries in the engine and performs the required clean up for each query.
  • Deletes the command topic for the cluster.

Example request

POST /ksql/terminate HTTP/1.1
Accept: application/vnd.ksql.v1+json
Content-Type: application/vnd.ksql.v1+json

{}

You can customize the clean up process if you want to delete some or all of the Kafka topics too:

Provide a List of Kafka Topics to Delete

You can provide a list of kafka topic names or regular expressions for Kafka topic names along with your TERMINATE CLUSTER request. The ksqlDB Server will delete all topics with names that are in the list or that match any of the regular expressions in the list. Only the topics that were generated by SQL queries are considered for deletion. Topics that were not generated by SQL queries aren’t deleted, even if they match the provided list. The following example shows how to delete topic FOO, along with all topics with the prefix bar.

POST /ksql/terminate HTTP/1.1
Accept: application/vnd.ksql.v1+json
Content-Type: application/vnd.ksql.v1+json

{
  "deleteTopicList": ["FOO", "bar.*"]
}