Skip hbo stats recording for nodes with dynamic filter #22853
Merged
Conversation
This file contains 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
feilong-liu
force-pushed
the
hbo_dynamic
branch
4 times, most recently
from
9aa48b4
to
49a1232
Compare
rschlussel
approved these changes
Jun 4, 2024
There was a problem hiding this comment.
What about dynamic filter for broadcast joins in Presto Java? 4332408
xiaoxmeng
reviewed
Jun 5, 2024
presto-tests/src/test/java/com/facebook/presto/execution/TestHistoryBasedStatsTracking.java
Show resolved
Hide resolved
| Optional<DynamicFilterStats> optionalDynamicFilterStats = Optional.empty(); | ||
| if (stats1.isPresent()) { | ||
| DynamicFilterStats dynamicFilterStats = stats1.get(); | ||
| stats2.ifPresent(dynamicFilterStats::mergeWith); |
There was a problem hiding this comment.
What does stats2.ifPresent(dynamicFilterStats::mergeWith) do here? Thanks!
There was a problem hiding this comment.
It's to merge the stats of the same plan node reported by different tasks
| } | ||
| } | ||
| } | ||
| Set<PlanNodeId> planNodeIdsDynamicFilter = new HashSet<>(); |
There was a problem hiding this comment.
s/planNodeIdsDynamicFilter/planNodeIdsWithDynamicFilterApplied/
...tion/src/test/java/com/facebook/presto/nativeworker/TestNativeHistoryBasedStatsTracking.java
Show resolved
Hide resolved
xiaoxmeng
approved these changes
Jun 5, 2024
36 tasks
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
Description
Presto CPP enable dynamic filter pushdown from join build to join probe side. When this is enabled, the number of output rows reported in the probe side will be less than the number of rows without dynamic filter pushdown.
This can be a problem if we still record the number of output rows in HBO. For example, for join T1 Join T2, after dynamic filter pushdown, it's possible that the probe side outputs 0 rows, and HBO will record 0 rows for probe side. Next time, we ran this query, HBO will use T1 as build side, as it's smaller. However now build side does not have dynamic filter pushdown, the build side may OOM due to no dynamic filter pushdown on build side.
In this PR, I will skip the HBO stats recording if it's affected by dynamic filter pushdown. It includes all plan nodes between the node where filter was pushed to (scan node in above example) and before the join node.
Motivation and Context
Described above
Impact
Avoid potential OOM due to inaccurate statistics caused by dynamic filter
Test Plan
End to end test
Control vs. Test
and unit test
Contributor checklist
Release Notes
Please follow release notes guidelines and fill in the release notes below.