Details
-
New Feature
-
Status: Resolved
-
Major
-
Resolution: Fixed
-
4.0.0
Description
State Store has been a black box from the introduction of the stateful operator. It has been the “internal” data to the streaming query, and Spark does not expose the data outside of the streaming query. There is no feature/tool for users to read and modify the content of state stores.
Specific to the ability to read the state, the lack of feature brings up various limitations like following:
- Users are unable to see the content in the state store, leading to inability to debug.
- Users have to perform some indirect approach on verifying the content of the state store in unit tests. The only option they can take is relying on the output of the query.
Given that, we propose to introduce a feature which enables users to read the state from the outside of the streaming query.
SPIP: https://docs.google.com/document/d/1_iVf_CIu2RZd3yWWF6KoRNlBiz5NbSIK0yThqG0EvPY/edit?usp=sharing
Attachments
Issue Links
- is depended upon by
-
SPARK-45671 Implement an option similar to corrupt record column in State Data Source Reader
- Open
-
SPARK-45672 Provide a unified user-facing schema for state format versions in state data source - reader
- Open
-
SPARK-45889 Implement push-down filter with partition ID and grouping key (if possible) for state data source reader
- Open
-
SPARK-45890 Implement limit push down for state data source reader
- Open
-
SPARK-45833 Document state data source - reader
- Resolved
-
SPARK-45888 Apply error class framework to state data source & state metadata data source
- Resolved
- links to