Memory Sink Connector

updatable

The Memory sink connector publishes query results to internal in-memory topics consumed by the Memory source.

Multiple sinks can publish to the same topic, and multiple sources can subscribe to the same topic. Use memory sinks and sources to construct Rule Pipelines.

Configuration Properties

Property Name	Optional	Description
topic	False	Target in-memory topic path (for example, analysis/result). Supports dynamic templates (such as {{.topic}}).
rowkindField	True	Field name specifying the action command (insert, update, upsert, or delete). Default is insert.
keyField	True	Primary key field used for table index updates in updatable sink configurations.

The Memory sink supports all common sink properties.

Example Configurations

Static topic destination:

{
  "memory": {
    "topic": "devices/result"
  }
}

Dynamic topic destination:

{
  "memory": {
    "topic": "{{.topic}}"
  }
}

Data Templates in Memory Sinks

Memory sinks transfer data objects directly in memory without serialization or deserialization. The sink ignores general format properties.

If you specify dataTemplate, the template must produce a valid JSON object string (for example, {"key": "{{.key}}"}). Array strings or raw non-JSON text are not supported.

Updatable Memory Sinks

The Memory sink functions as an Updatable Sink. Updatable memory sinks modify the state of in-memory lookup tables subscribed to the same topic.

In this example, the rule writes state updates to the in-memory topic alertVal. The action field specifies the operation verb (upsert, delete), and id serves as the primary key:

{
  "id": "ruleUpdateAlert",
  "sql": "SELECT * FROM alertStream",
  "actions": [
    {
      "memory": {
        "keyField": "id",
        "rowkindField": "action",
        "topic": "alertVal",
        "sendSingle": true
      }
    }
  ]
}