Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[BugFix] Clear probe RF whose probe expr contains dict mapping expr (backport #50690) #50716

Closed
wants to merge 1 commit into from

Conversation

mergify[bot]
Copy link
Contributor

@mergify mergify bot commented Sep 5, 2024

Why I'm doing:

If the equi-join predicate is derived from an expression evaluated on a column with low cardinality, a corresponding runtime filter will be generated on the OlapScanNode. The probe expression for this filter is a DictMappingExpr.

Except for the main branch, this DictMappingExpr in the probe expression is not rewritten, leading to a crash when evaluated using a DictID column.

*** Aborted at 1725338148 (unix time) try "date -d @1725338148" if you are using GNU date ***
PC: @          0x2dc0ce0 starrocks::memequal()
*** SIGSEGV (@0x7) received by PID 167210 (TID 0x7f57b5d00700) from PID 7; stack trace: ***
    @          0x5c32442 google::(anonymous namespace)::FailureSignalHandler()
    @     0x7f58f08c7630 (unknown)
    @          0x2dc0ce0 starrocks::memequal()
    @          0x3a6b20a starrocks::vectorized::UnpackConstColumnBinaryFunction<>::evaluate<>()
    @          0x3a6b61c starrocks::vectorized::VectorizedBinaryPredicate<>::evaluate_checked()
    @          0x4045b19 starrocks::vectorized::DictMappingExpr::evaluate_checked()
    @          0x3b7df24 starrocks::vectorized::VectorizedCaseExpr<>::evaluate_no_case()
    @          0x3b84e0d starrocks::vectorized::VectorizedCaseExpr<>::evaluate_checked()
    @          0x394a3b3 starrocks::ExprContext::evaluate()
    @          0x4133af0 starrocks::vectorized::RuntimeFilterProbeCollector::update_selectivity()
    @          0x4134caf starrocks::vectorized::RuntimeFilterProbeCollector::evaluate()
    @          0x2dd4506 starrocks::pipeline::Operator::eval_runtime_bloom_filters()
    @          0x2ddab55 starrocks::pipeline::ScanOperator::pull_chunk()
    @          0x2db7ed0 starrocks::pipeline::PipelineDriver::process()
    @          0x5207fca starrocks::pipeline::GlobalDriverExecutor::_worker_thread()
    @          0x4bea3a2 starrocks::ThreadPool::dispatch_thread()
    @          0x4be4e3a starrocks::Thread::supervise_thread()
    @     0x7f58f08bfea5 start_thread
    @     0x7f58efedab0d __clone
    @                0x0 (unknown)

Plan:

  3:Decode
  |  <dict id 38> : <string id 2>
  |  cardinality: 1
  |  probe runtime filters:
  |  - filter_id = 1, probe_expr = (2: P_NAME)
  |  column statistics: 
  |  * P_NAME-->[-Infinity, Infinity, 0.0, 1.0, 1.0] UNKNOWN
  |  * P_BRAND-->[-Infinity, Infinity, 0.0, 1.0, 1.0] UNKNOWN
  |  * cast-->[-Infinity, Infinity, 0.0, 16.0, 3.0] ESTIMATE
  |  
  2:Project
  |  output columns:
  |  12 <-> CASE WHEN DictDecode(38: P_NAME, [<place-holder> = 'a']) THEN 'a1' WHEN DictDecode(39: P_BRAND, [<place-holder> = 'b']) THEN 'b1' ELSE 'c1' END
  |  38 <-> [38: P_NAME, INT, false]
  |  cardinality: 1
  |  column statistics: 
  |  * cast-->[-Infinity, Infinity, 0.0, 16.0, 3.0] ESTIMATE
  |  
  1:OlapScanNode
     table: part_v2, rollup: part_v2
     preAggregation: on
     dict_col=P_NAME,P_BRAND
     partitionsRatio=1/1, tabletsRatio=10/10
     tabletList=14325,14327,14329,14331,14333,14335,14337,14339,14341,14343
     actualRows=0, avgRowSize=18.0
     cardinality: 1
     probe runtime filters:
     - filter_id = 0, probe_expr = (CASE WHEN DictDecode(38: P_NAME, [<place-holder> = 'a']) THEN 'a1' WHEN DictDecode(39: P_BRAND, [<place-holder> = 'b']) THEN 'b1' ELSE 'c1' END)

What I'm doing:

To address this case, a check is performed in the ProjectNode. If an expression contains a DictMappingExpr, the RF will be no longer pushed down to its children.

Additionally, both FE and BE have implemented a fallback strategy to remove all probe runtime filters that include a DictMappingExpr in the probe expression.

Plan after this PR:

  3:Decode
  |  <dict id 38> : <string id 2>
  |  cardinality: 1
  |  probe runtime filters:
  |  - filter_id = 1, probe_expr = (2: P_NAME)
  |  column statistics: 
  |  * P_NAME-->[-Infinity, Infinity, 0.0, 1.0, 1.0] UNKNOWN
  |  * P_BRAND-->[-Infinity, Infinity, 0.0, 1.0, 1.0] UNKNOWN
  |  * cast-->[-Infinity, Infinity, 0.0, 16.0, 3.0] ESTIMATE
  |  
  2:Project
  |  output columns:
  |  12 <-> CASE WHEN DictDecode(38: P_NAME, [<place-holder> = 'a']) THEN 'a1' WHEN DictDecode(39: P_BRAND, [<place-holder> = 'b']) THEN 'b1' ELSE 'c1' END
  |  38 <-> [38: P_NAME, INT, false]
  |  cardinality: 1
  |  probe runtime filters:
  |  - filter_id = 0, probe_expr = (<slot 12>)
  |  column statistics: 
  |  * cast-->[-Infinity, Infinity, 0.0, 16.0, 3.0] ESTIMATE
  |  
  1:OlapScanNode
     table: part_v2, rollup: part_v2
     preAggregation: on
     dict_col=P_NAME,P_BRAND
     partitionsRatio=0/1, tabletsRatio=0/0
     tabletList=
     actualRows=0, avgRowSize=18.0
     cardinality: 1
     column statistics: 
     * P_NAME-->[-Infinity, Infinity, 0.0, 1.0, 1.0] UNKNOWN
     * P_BRAND-->[-Infinity, Infinity, 0.0, 1.0, 1.0] UNKNOWN
     * cast-->[-Infinity, Infinity, 0.0, 16.0, 3.0] ESTIMATE
     * 

Fixes #issue

What type of PR is this:

  • BugFix
  • Feature
  • Enhancement
  • Refactor
  • UT
  • Doc
  • Tool

Does this PR entail a change in behavior?

  • Yes, this PR will result in a change in behavior.
  • No, this PR will not result in a change in behavior.

If yes, please specify the type of change:

  • Interface/UI changes: syntax, type conversion, expression evaluation, display information
  • Parameter changes: default values, similar parameters but with different default values
  • Policy changes: use new policy to replace old one, functionality automatically enabled
  • Feature removed
  • Miscellaneous: upgrade & downgrade compatibility, etc.

Checklist:

  • I have added test cases for my bug fix or my new feature
  • This pr needs user documentation (for new or modified features or behaviors)
    • I have added documentation for my new feature or new function
  • This is a backport pr

Bugfix cherry-pick branch check:

  • I have checked the version labels which the pr will be auto-backported to the target branch
    • 3.3
    • 3.2
    • 3.1
    • 3.0
    • 2.5

This is an automatic backport of pull request #50690 done by [Mergify](https://mergify.com). ## Why I'm doing:

If the equi-join predicate is derived from an expression evaluated on a column with low cardinality, a corresponding runtime filter will be generated on the OlapScanNode. The probe expression for this filter is a DictMappingExpr.

Except for the main branch, this DictMappingExpr in the probe expression is not rewritten, leading to a crash when evaluated using a DictID column.

*** Aborted at 1725338148 (unix time) try "date -d @1725338148" if you are using GNU date ***
PC: @          0x2dc0ce0 starrocks::memequal()
*** SIGSEGV (@0x7) received by PID 167210 (TID 0x7f57b5d00700) from PID 7; stack trace: ***
    @          0x5c32442 google::(anonymous namespace)::FailureSignalHandler()
    @     0x7f58f08c7630 (unknown)
    @          0x2dc0ce0 starrocks::memequal()
    @          0x3a6b20a starrocks::vectorized::UnpackConstColumnBinaryFunction<>::evaluate<>()
    @          0x3a6b61c starrocks::vectorized::VectorizedBinaryPredicate<>::evaluate_checked()
    @          0x4045b19 starrocks::vectorized::DictMappingExpr::evaluate_checked()
    @          0x3b7df24 starrocks::vectorized::VectorizedCaseExpr<>::evaluate_no_case()
    @          0x3b84e0d starrocks::vectorized::VectorizedCaseExpr<>::evaluate_checked()
    @          0x394a3b3 starrocks::ExprContext::evaluate()
    @          0x4133af0 starrocks::vectorized::RuntimeFilterProbeCollector::update_selectivity()
    @          0x4134caf starrocks::vectorized::RuntimeFilterProbeCollector::evaluate()
    @          0x2dd4506 starrocks::pipeline::Operator::eval_runtime_bloom_filters()
    @          0x2ddab55 starrocks::pipeline::ScanOperator::pull_chunk()
    @          0x2db7ed0 starrocks::pipeline::PipelineDriver::process()
    @          0x5207fca starrocks::pipeline::GlobalDriverExecutor::_worker_thread()
    @          0x4bea3a2 starrocks::ThreadPool::dispatch_thread()
    @          0x4be4e3a starrocks::Thread::supervise_thread()
    @     0x7f58f08bfea5 start_thread
    @     0x7f58efedab0d __clone
    @                0x0 (unknown)

Plan:

  3:Decode
  |  <dict id 38> : <string id 2>
  |  cardinality: 1
  |  probe runtime filters:
  |  - filter_id = 1, probe_expr = (2: P_NAME)
  |  column statistics: 
  |  * P_NAME-->[-Infinity, Infinity, 0.0, 1.0, 1.0] UNKNOWN
  |  * P_BRAND-->[-Infinity, Infinity, 0.0, 1.0, 1.0] UNKNOWN
  |  * cast-->[-Infinity, Infinity, 0.0, 16.0, 3.0] ESTIMATE
  |  
  2:Project
  |  output columns:
  |  12 <-> CASE WHEN DictDecode(38: P_NAME, [<place-holder> = 'a']) THEN 'a1' WHEN DictDecode(39: P_BRAND, [<place-holder> = 'b']) THEN 'b1' ELSE 'c1' END
  |  38 <-> [38: P_NAME, INT, false]
  |  cardinality: 1
  |  column statistics: 
  |  * cast-->[-Infinity, Infinity, 0.0, 16.0, 3.0] ESTIMATE
  |  
  1:OlapScanNode
     table: part_v2, rollup: part_v2
     preAggregation: on
     dict_col=P_NAME,P_BRAND
     partitionsRatio=1/1, tabletsRatio=10/10
     tabletList=14325,14327,14329,14331,14333,14335,14337,14339,14341,14343
     actualRows=0, avgRowSize=18.0
     cardinality: 1
     probe runtime filters:
     - filter_id = 0, probe_expr = (CASE WHEN DictDecode(38: P_NAME, [<place-holder> = 'a']) THEN 'a1' WHEN DictDecode(39: P_BRAND, [<place-holder> = 'b']) THEN 'b1' ELSE 'c1' END)

What I'm doing:

To address this case, a check is performed in the ProjectNode. If an expression contains a DictMappingExpr, the RF will be no longer pushed down to its children.

Additionally, both FE and BE have implemented a fallback strategy to remove all probe runtime filters that include a DictMappingExpr in the probe expression.

Plan after this PR:

  3:Decode
  |  <dict id 38> : <string id 2>
  |  cardinality: 1
  |  probe runtime filters:
  |  - filter_id = 1, probe_expr = (2: P_NAME)
  |  column statistics: 
  |  * P_NAME-->[-Infinity, Infinity, 0.0, 1.0, 1.0] UNKNOWN
  |  * P_BRAND-->[-Infinity, Infinity, 0.0, 1.0, 1.0] UNKNOWN
  |  * cast-->[-Infinity, Infinity, 0.0, 16.0, 3.0] ESTIMATE
  |  
  2:Project
  |  output columns:
  |  12 <-> CASE WHEN DictDecode(38: P_NAME, [<place-holder> = 'a']) THEN 'a1' WHEN DictDecode(39: P_BRAND, [<place-holder> = 'b']) THEN 'b1' ELSE 'c1' END
  |  38 <-> [38: P_NAME, INT, false]
  |  cardinality: 1
  |  probe runtime filters:
  |  - filter_id = 0, probe_expr = (<slot 12>)
  |  column statistics: 
  |  * cast-->[-Infinity, Infinity, 0.0, 16.0, 3.0] ESTIMATE
  |  
  1:OlapScanNode
     table: part_v2, rollup: part_v2
     preAggregation: on
     dict_col=P_NAME,P_BRAND
     partitionsRatio=0/1, tabletsRatio=0/0
     tabletList=
     actualRows=0, avgRowSize=18.0
     cardinality: 1
     column statistics: 
     * P_NAME-->[-Infinity, Infinity, 0.0, 1.0, 1.0] UNKNOWN
     * P_BRAND-->[-Infinity, Infinity, 0.0, 1.0, 1.0] UNKNOWN
     * cast-->[-Infinity, Infinity, 0.0, 16.0, 3.0] ESTIMATE
     * 

Fixes #issue

What type of PR is this:

  • BugFix
  • Feature
  • Enhancement
  • Refactor
  • UT
  • Doc
  • Tool

Does this PR entail a change in behavior?

  • Yes, this PR will result in a change in behavior.
  • No, this PR will not result in a change in behavior.

If yes, please specify the type of change:

  • Interface/UI changes: syntax, type conversion, expression evaluation, display information
  • Parameter changes: default values, similar parameters but with different default values
  • Policy changes: use new policy to replace old one, functionality automatically enabled
  • Feature removed
  • Miscellaneous: upgrade & downgrade compatibility, etc.

Checklist:

  • I have added test cases for my bug fix or my new feature
  • This pr needs user documentation (for new or modified features or behaviors)
    • I have added documentation for my new feature or new function
  • This is a backport pr

…50690)

Signed-off-by: zihe.liu <[email protected]>
(cherry picked from commit a810948)

# Conflicts:
#	be/src/exprs/vectorized/runtime_filter_bank.cpp
#	fe/fe-core/src/main/java/com/starrocks/planner/ProjectNode.java
@mergify mergify bot added the conflicts label Sep 5, 2024
Copy link
Contributor Author

mergify bot commented Sep 5, 2024

Cherry-pick of a810948 has failed:

On branch mergify/bp/branch-2.5/pr-50690
Your branch is up to date with 'origin/branch-2.5'.

You are currently cherry-picking commit a810948f01.
  (fix conflicts and run "git cherry-pick --continue")
  (use "git cherry-pick --skip" to skip this patch)
  (use "git cherry-pick --abort" to cancel the cherry-pick operation)

Changes to be committed:
	modified:   fe/fe-core/src/main/java/com/starrocks/analysis/Expr.java
	modified:   fe/fe-core/src/main/java/com/starrocks/planner/PlanFragment.java
	modified:   fe/fe-core/src/main/java/com/starrocks/sql/plan/PlanFragmentBuilder.java
	modified:   fe/fe-core/src/test/java/com/starrocks/sql/plan/LowCardinalityTest.java

Unmerged paths:
  (use "git add <file>..." to mark resolution)
	both modified:   be/src/exprs/vectorized/runtime_filter_bank.cpp
	both modified:   fe/fe-core/src/main/java/com/starrocks/planner/ProjectNode.java

To fix up this pull request, you can check it out locally. See documentation: https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/reviewing-changes-in-pull-requests/checking-out-pull-requests-locally

Copy link
Contributor Author

mergify bot commented Sep 5, 2024

@mergify[bot]: Backport conflict, please reslove the conflict and resubmit the pr

@mergify mergify bot deleted the mergify/bp/branch-2.5/pr-50690 branch September 5, 2024 02:21
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant