kazuyukitanimura commented on code in PR #7400:
URL: https://github.com/apache/arrow-datafusion/pull/7400#discussion_r1326550421
##########
datafusion/core/src/physical_plan/aggregates/row_hash.rs:
##########
@@ -466,15 +625,122 @@ impl GroupedHashAggregateStream {
for acc in self.accumulators.iter_mut() {
match self.mode {
AggregateMode::Partial => output.extend(acc.state(emit_to)?),
+ _ if spilling => {
+ // If spilling, output partial state because the spilled
data will be
+ // merged and re-evaluated later.
+ output.extend(acc.state(emit_to)?)
+ }
AggregateMode::Final
| AggregateMode::FinalPartitioned
| AggregateMode::Single
| AggregateMode::SinglePartitioned =>
output.push(acc.evaluate(emit_to)?),
}
}
- self.update_memory_reservation()?;
- let batch = RecordBatch::try_new(self.schema(), output)?;
+ // emit reduces the memory usage. Ignore Err from
update_memory_reservation. Even if it is
+ // over the target memory size after emission, we can emit again
rather than returning Err.
+ let _ = self.update_memory_reservation();
+ let batch = RecordBatch::try_new(schema, output)?;
Ok(batch)
}
+
+ /// Optimistically, [`Self::group_aggregate_batch`] allows to exceed the
memory target slightly
+ /// (~ 1 [`RecordBatch`]) for simplicity. In such cases, spill the data to
disk and clear the
+ /// memory. Currently only [`GroupOrdering::None`] is supported for
spilling.
+ fn spill_previous_if_necessary(&mut self, batch: &RecordBatch) ->
Result<()> {
+ // TODO: support group_ordering for spilling
+ if self.group_values.len() > 0
+ && batch.num_rows() > 0
+ && matches!(self.group_ordering, GroupOrdering::None)
+ && !matches!(self.mode, AggregateMode::Partial)
+ && !self.spill_state.is_stream_merging
+ && self.update_memory_reservation().is_err()
+ {
+ // Use input batch (Partial mode) schema for spilling because
+ // the spilled data will be merged and re-evaluated later.
+ self.spill_state.spill_schema = batch.schema();
+ self.spill()?;
+ self.clear_shrink(batch);
+ }
+ Ok(())
+ }
+
+ /// Emit all rows, sort them, and store them on disk.
+ fn spill(&mut self) -> Result<()> {
+ let emit = self.emit(EmitTo::All, true)?;
+ let sorted = sort_batch(&emit, &self.spill_state.spill_expr, None)?;
Review Comment:
> ... so they are continuously aggregated into final aggregation result.
During this re-grouping time, we may run out of memory again. The only
option is to streaming read and early outputting. In order to do so, the input
for regrouping has to be sorted. Otherwise we cannot output the final
aggregation results early and that makes the hash table keep all rows in
memory. `a = 2` can be at very first and very last. So the very last step has
to be a sort based aggregation in order to avoid OOM.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
