pushdown:group-unsupported: the {adapter} ad...

Warnings reference › source.pushdown.group-unsupported.*.*

Identifier

source.pushdown.group-unsupported.*.*

Raised by

The grid (core)

Since

1.79.0

What happened

The grid is grouped over a pushdown source and the engine was not asked to group, for the reason named: the adapter declares neither capabilities.group nor the statistics to build the level from grouped aggregates (Splunk, REST, OData, GraphQL, DFQL), or part of this query did not push, so the engine would group the wrong set. A remote source holds a window of leaves, and group rows built from it would show one page's figures as a group's, so the grid stays flat: every display row is a leaf, grid.rows.count() is the source's count and no row carries a subtotal. The next query the engine can group is shown grouped again.

The fix

Group over an adapter that declares capabilities.group (DuckDB) or builds groups from grouped aggregates (ClickHouse, Elasticsearch), or implement executeGroupLevel or executeGroupedAggregates on your own; where the reason is a filter, sort or quick search that did not push, narrow the query to what the adapter takes. To group client-side over every matching row, set fullDataset: { enabled: true } within its memory limits. lastPlan().groupReason carries the reason programmatically.

Reference

Covered in full at the API reference.