2010YOUY01 commented on code in PR #23838: URL: https://github.com/apache/datafusion/pull/23838#discussion_r3643705212
########## datafusion/sqllogictest/test_files/aggregate_memory_limit.slt: ########## @@ -0,0 +1,137 @@ +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license agreements. See the NOTICE file +# distributed with this work for additional information +# regarding copyright ownership. The ASF licenses this file +# to you under the Apache License, Version 2.0 (the +# "License"); you may not use this file except in compliance +# with the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an +# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +# KIND, either express or implied. See the License for the +# specific language governing permissions and limitations +# under the License. + +# Memory-limited (spilling) grouped hash aggregation. +# +# High-cardinality GROUP BY under a tight memory limit: the aggregate spills to +# disk, re-groups the spilled state, and must still return the right answer. +# +# The group key is scrambled with `(v * 7) % 100000` because generate_series is +# sorted, which would take the streaming path that never spills. gcd(7, 100000) +# = 1, so it's a bijection over 1..100000 -- still 100000 groups, just unsorted, +# so the hash table grows and spills. +# +# Each query aggregates over the grouped result, so the expected output is one +# row. sum(1..100000) = 5000050000, and every v lands in one group, so the +# per-group sums always add back to that total. + +# Single partition keeps the aggregation in one operator (no repartition). +statement ok +SET datafusion.execution.target_partitions = 1 + +statement ok +SET datafusion.runtime.memory_limit = '1M' + +# --- Case A: single-column high-cardinality GROUP BY --- +query II +SELECT count(*), sum(total) +FROM ( + SELECT (v * 7) % 100000 AS k, sum(v) AS total + FROM generate_series(1, 100000) AS t(v) + GROUP BY (v * 7) % 100000 +) +---- +100000 5000050000 + +# Prove the inner aggregate actually spills (else these tests would silently stop covering the spill path). +# Only `spill_count` is pinned; the other metrics vary per run. Review Comment: Let's assert the `spill_count` similarly in other queries. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
