[SPARK-45558] Introduce a metadata file for streaming stateful operator

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.5.0
• Fix Version/s: 4.0.0
• Component/s: Structured Streaming
• Labels: pull-request-available
• Assignee: Chaoqin Li
• Reporter: Chaoqin Li
• Created: 16/Oct/23 17:23
• Updated: 18/Oct/23 06:49
• Resolved: 18/Oct/23 06:49

Description

The information to store in the metadata file:

• operator name (no need to be unique among stateful operators in the query)

• state store name
• numColumnsPrefixKey: > 0 if prefix scan is enabled, 0 otherwise

The body of metadata file will be in json format. The metadata file will be versioned just as other streaming metadata file to be future proof.

The metadata file will improve expose more information about the state store, improves debugability and facilitate the development of state related feature such as reading and writing state and state repartitioning.

Issue Links

• links to: GitHub Pull Request #43393