Sink types

Note: Types are just Scala classes that translates on JSON directly, for example, for class case class Buzz(x: Int, y: Map[String, String]) valid JSON representation could be: { "x": 120, "y": {"a": 1} }. Also Option types represent optional parameters.

Generic parameters:

• rowSchema (JSON object) is the row schema for sink (keys are incidents fields, values are table columns/JSON object keys). All fields are required and have a string type unless other specified.
  • fromTsField - the column name for the starting timestamp of an incident
  • toTsField - the column name for the ending timestamp of an incident
  • unitIdField - the column name for the unit ID of an event (the value is taken from source)
  • appIdFieldVal (string and integer) - the column name and value for the incident type (for example, to distinguish different incident categories).
  • patternIdField

Example

{
  "toTsField": "to_ts",
  "fromTsField": "from_ts",
  "unitIdField": "engine_id",
  "appIdFieldVal": ["rule_type", 1],
  "patternIdField": "rule_id",
  "subunitIdField": "physical_id",
  "incidentIdField": "uuid"
}

JDBC sink

• jdbcUrl, driverName, userName, password - the same as in JDBC source
• tableName (string) is the name of an SQL table to store incidents.
• batchInterval (optional, integer) is the size of a single batch to write (in rows)

Example

(no rowSchema is given, see above)

{
  "jdbcUrl": "jdbc:clickhouse://default:@127.0.0.1:8123/mydb",
  "tableName": "engine_events",
  "driverName": "com.clickhouse.jdbc.ClickHouseDriver"
}

Kafka sink

• broker and topic - the same of Kafka sink (note: for historical reasons, it is called broker in singular in sink)
• serializer (string, default "json") is the serializer for incident messages.

Example

(no rowSchema is given, see above)

{
  "broker": "10.83.0.3:9092",
  "topic": "engine_events"
}