Skip to content

[FLINK-40324][table] Append-only should stay unmaterialized without an ON CONFLICT clause - #28918

Open
gustavodemorais wants to merge 1 commit into
apache:masterfrom
confluentinc:FLINK-40324
Open

[FLINK-40324][table] Append-only should stay unmaterialized without an ON CONFLICT clause#28918
gustavodemorais wants to merge 1 commit into
apache:masterfrom
confluentinc:FLINK-40324

Conversation

@gustavodemorais

Copy link
Copy Markdown
Contributor

What is the purpose of the change

An append-only query into a sink with a primary key and no ON CONFLICT clause currently enables the sink upsert materializer, adding a SinkUpsertMaterializer and a keyed shuffle the pipeline did not have before FLIP-558. This keeps those queries unmaterialized, since an absent clause selects DEDUPLICATE and insert-only input has nothing to deduplicate. It is only reachable with table.exec.sink.require-on-conflict set to false, which otherwise rejects the missing clause.

Brief change log

  • Short-circuit the AUTO upsert-materialize branch for insert-only input when no ON CONFLICT clause is specified.
  • Move the short-circuit after the require-on-conflict validation so a missing clause is still reported.
  • Extract validateOnConflictSpecifiedIfRequired out of analyzeUpsertMaterializeStrategy.

Verifying this change

  • TableSinkTest

Does this pull request potentially affect one of the following parts:

  • Dependencies (does it add or upgrade a dependency): no
  • The public API, i.e., is any changed class annotated with @Public(Evolving): no
  • The serializers: no
  • The runtime per-record code paths (performance sensitive): no
  • Anything that affects deployment or recovery: JobManager (and its components), Checkpointing, Kubernetes/Yarn, ZooKeeper: yes - a re-planned job drops SinkUpsertMaterializer and its keyed state, so the topology changes on restore; compiled plans are unaffected because requireUpsertMaterialize is persisted
  • The S3 file system connector: no

Documentation

  • Does this pull request introduce a new feature? no
  • If yes, how is the feature documented? not applicable

Was generative AI tooling used to co-author this PR?
  • Yes (please specify the tool below)

2.1.220 (Claude Code) with Opus 5

@flinkbot

flinkbot commented Aug 4, 2026

Copy link
Copy Markdown
Collaborator

CI report:

Bot commands The @flinkbot bot supports the following commands:
  • @flinkbot run azure re-run the last Azure build

@twalthr twalthr left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

One minor comment, otherwise LGTM

sink.conflictStrategy == null || sink.isDeduplicateConflictStrategy

// For a DEDUPLICATE strategy and INSERT only input, we simply let the inserts be handled
// as UPSERT_AFTER and overwrite previous value

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
// as UPSERT_AFTER and overwrite previous value
// as UPDATE_AFTER and overwrite previous value

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ah, old comment has this and we just copied. Thanks, fixed

…n ON CONFLICT clause

Before FLIP-558, insert-only input never enabled the sink upsert materializer.
The AUTO branch now only short-circuits when DEDUPLICATE is stated explicitly,
so an append-only query into a sink with a primary key and no ON CONFLICT
clause enables materialization instead, adding a SinkUpsertMaterializer and a
keyed shuffle the pipeline did not have before.

Omitting the clause selects DEDUPLICATE, so it has to short-circuit as well.
The check moves after the require-on-conflict validation so that enforcement
still reports a missing clause.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants