This documentation is for an unreleased version of Apache Flink Stateful Functions. We recommend you use the latest stable version.
Configurations
Configurations #
Stateful Functions includes a small number of SDK specific configurations.
These may be set through your job’s flink-conf.yaml
.
Key | Default | Type | Description |
---|---|---|---|
statefun.module.global-config.<KEY> |
(none) | String | Adds the given key/value pair to the Stateful Functions global configuration. |
statefun.message.serializer |
WITH_PROTOBUF_PAYLOADS | Message Serializer | The serializer to use for on the wire messages. Options are WITH_PROTOBUF_PAYLOADS, WITH_KRYO_PAYLOADS, WITH_RAW_PAYLOADS, WITH_CUSTOM_PAYLOADS. |
statefun.message.custom-payload-serializer-class |
(none) | String | The custom payload serializer class to use with the WITH_CUSTOM_PAYLOADS serializer, which must implement MessagePayloadSerializer. |
statefun.flink-job-name |
StatefulFunctions | String | The name to display in the Flink-UI. |
statefun.feedback.memory.size |
32 MB | Memory | The number of bytes to use for in memory buffering of the feedback channel, before spilling to disk. |
statefun.async.max-per-task |
1024 | Integer | The max number of async operations per task before backpressure is applied. |
statefun.embedded |
false | Boolean | Set to 'true' if Flink is running this job from an uber jar, rather than using statefun-specific docker images. This disables the validation of whether 'classloader.parent-first-patterns.additional' contains 'org.apache.flink.statefun', 'org.apache.kafka' and 'com.google.protobuf' patterns. It is then up to the creator of the uber jar to ensure that the three dependencies (statefun, kafka and protobuf) don't have version conflicts. |