[SPARK-36594][SQL] ORC vectorized reader should properly check maximal number of fields #33842
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Contributor
Author
|
cc @cloud-fan and @dongjoon-hyun could you help take a look when you have time? Thanks. |
Contributor
Author
|
This cannot be merged into 3.2 branch cleanly as it depends on #33626 which was merged only on master. Will create another PR for 3.2 branch once this looks good. |
cloud-fan
approved these changes
Aug 26, 2021
|
Kubernetes integration test starting |
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/orc/OrcFileFormat.scala
Show resolved
Hide resolved
gengliangwang
approved these changes
Aug 26, 2021
|
Kubernetes integration test status failure |
cloud-fan
pushed a commit
that referenced
this pull request
Aug 26, 2021
…aximal number of fields ### What changes were proposed in this pull request? This is the patch on branch-3.2 for #33842. See the description in the other PR. ### Why are the changes needed? Avoid OOM/performance regression when reading ORC table with nested column types. ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? Added unit test in `OrcSourceSuite.scala`. Closes #33843 from c21/branch-3.2. Authored-by: Cheng Su <[email protected]> Signed-off-by: Wenchen Fan <[email protected]>
Member
|
Merging to master |
|
Test build #142780 has finished for PR 33842 at commit
|
Contributor
Author
|
Thank you @cloud-fan and @gengliangwang for review! |
Member
|
+1, LGTM. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
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Outdated suggestions cannot be applied.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.
What changes were proposed in this pull request?
Debugged internally and found a bug where we should disable vectorized reader now based on schema recursively. Currently we check
schema.lengthto be no more thanwholeStageMaxNumFieldsto enable vectorization.schema.lengthdoes not take nested columns sub-fields into condition (i.e. view nested column same as primitive column). This check will be wrong when enabling vectorization for nested columns. We should follow same check fromWholeStageCodegenExecto check sub-fields recursively. This will not cause correctness issue but will cause performance issue where we may enable vectorization for nested columns by mistake when nested column has a lot of sub-fields.Why are the changes needed?
Avoid OOM/performance regression when reading ORC table with nested column types.
Does this PR introduce any user-facing change?
No.
How was this patch tested?
Added unit test in
OrcQuerySuite.scala. Verified test failed without this change.