Few changes and fixes to cube's Materialization.

[agorajek]

Summary

Two main changes:

1. Don't bump node version when adding/refreshing materialization.
2. Add a new end point for materialization refreshes. Not critical, but nice to have.

Test Plan

Unit tests plus by hand locally.

• PR has an associated issue: #
• make check passes
• make test shows 100% unit test coverage

Deployment Plan

Auto.

[netlify]

✅ Deploy Preview for thriving-cassata-78ae72 canceled.

Name	Link
🔨 Latest commit	b5c601e
🔍 Latest deploy log	https://app.netlify.com/projects/thriving-cassata-78ae72/deploys/699d2f6e0647e00008c824ff

[Copilot]

Pull request overview

This PR updates cube materialization “rebuild/refresh” behavior so workflows can be regenerated without creating a new cube version, and introduces a dedicated query-service endpoint for that refresh operation.

Changes:

• Update cube node update logic to refresh materialization workflows without bumping the cube version when no other cube definition changes are present.
• Add a refresh_cube_materialization client method and wire it through the HTTP query client wrapper.
• Update UI wording/logic around “Rebuild (latest) Materialization” and refresh button visibility; update dependencies/lockfiles.

Reviewed changes

Copilot reviewed 8 out of 9 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
uv.lock	Dependency lockfile updates / marker simplification.
datajunction-ui/src/app/pages/NodePage/NodeMaterializationTab.jsx	Adjust tab selection defaults and update “rebuild” messaging + latest-materialization detection.
datajunction-server/datajunction_server/internal/nodes.py	Refresh materializations without version bump when refresh_materialization=true and no other cube changes.
datajunction-server/datajunction_server/service_clients.py	Add query-service client method to POST /cubes/{cube}/refresh-materialization.
datajunction-server/datajunction_server/query_clients/http.py	Delegate refresh call through the HTTP query client wrapper.
datajunction-server/tests/service_clients_test.py	Unit tests for the new service client refresh method.
datajunction-server/tests/query_clients/http_query_client_test.py	Unit test that wrapper delegates refresh call.
datajunction-server/tests/api/cubes_test.py	API test ensuring refresh doesn’t bump cube version and reactivates matching deactivated materializations.
datajunction-query/pyproject.toml	Remove psycopg[async] extra (keep pool).

Comments suppressed due to low confidence (1)

datajunction-server/datajunction_server/service_clients.py:592

• This POST to the query service does not specify a timeout. Other query-service calls (e.g., cube materialize/backfill/deactivate) set explicit timeouts, and leaving this unbounded risks hanging requests/workers if the query service is slow or unresponsive. Add an explicit timeout consistent with similar cube endpoints.

        response = self.requests_session.post(
            refresh_endpoint,
            headers=headers,
            params=params,
            json=body,
        )

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

refresh_materialization can be triggered even when query_service_client is a non-HTTP implementation (e.g., configured Snowflake client). In that case refresh_cube_materialization may not exist and this will raise an AttributeError at runtime. Consider guarding the call (e.g., hasattr(query_service_client, "refresh_cube_materialization") / isinstance(HttpQueryServiceClient, ...)) and returning a clear error, or add a default refresh_cube_materialization to the query-client interface with a controlled NotImplementedError and handle it here.

Suggested change

	query_service_client.refresh_cube_materialization(
	cube_name=node_revision.name,
	cube_version=node_revision.version,
	materializations=active_mats,
	request_headers=request_headers,
	)
	refresh_method = getattr(
	query_service_client,
	"refresh_cube_materialization",
	None,
	)
	if not callable(refresh_method):
	_logger.error(
	"Query service client %s does not support cube materialization "
	"refresh for cube=%s",
	type(query_service_client).__name__,
	node_revision.name,
	)
	raise DJInvalidInputException(
	"Refreshing cube materializations is not supported by the "
	"configured query service client.",
	)
	try:
	refresh_method(
	cube_name=node_revision.name,
	cube_version=node_revision.version,
	materializations=active_mats,
	request_headers=request_headers,
	)
	except NotImplementedError as exc: # pragma: no cover
	_logger.error(
	"Cube materialization refresh is not implemented by client %s "
	"for cube=%s",
	type(query_service_client).__name__,
	node_revision.name,
	)
	raise DJInvalidInputException(
	"Refreshing cube materializations is not supported by the "
	"configured query service client.",
	) from exc