Change AQE default minPartitionNum to defaultParallelism #674
+106
−62
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Applying changes from SPARK-31124 / apache/spark#27879.
What changes were proposed in this pull request?
Authoritative description on apache/spark#27879.
Summary: When coalescing partitions in AQE, the minimum is no longer 1 but default parallelism. The default parallelism is
spark.default.parallelism
when set or the number of available CPUs at any given point in time. This ensures that we don't lose parallelism by applying AQE.Why didn't you just take the upstream commit? The AQE framework was refactored in SPARK-31037, which in turn relies on changes we haven't picked yet. And the sum of those was more than I felt comfortable with taking in one batch. So yes this is lame. We can just revert this later.
How was this patch tested? I wrote a unit test which ensures that the partition number is the default parallelism value where it would otherwise be 1.