HIVE-29367: prevent Long overflows in ConvertJoinMapJoin #6237
+475
−11
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Outdated suggestions cannot be applied.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?
HIVE-29367: fixing overflows in ConvertJoinMapJoin calculations
Why are the changes needed?
ConvertJoinMapJoin does not use StatsUtils.safeAdd()/saveMult() for all its calculations. There are some real life scenarios when it could perform a catastrophic decision to convert a join to a mapjoin after calculating negative size for the 'small" table, resulting in an OOM during query processing
Does this PR introduce any user-facing change?
No
How was this patch tested?
Via unit testing and with load testing on a custom Hive installation based of 4.0x version
You can see the test output generated by the pre-fix code here:
it clearly confirms the decision of perform a mapjoin despite very large volume of data