Do not use SkewedPartitionRebalancer for remote exchange in FTE #21979
+7
−2
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
findepi
approved these changes
May 16, 2024
|
CI (re-running) |
sopel39
approved these changes
May 16, 2024
core/trino-main/src/main/java/io/trino/sql/planner/LocalExecutionPlanner.java
Outdated
Show resolved
Hide resolved
core/trino-main/src/main/java/io/trino/sql/planner/LocalExecutionPlanner.java
Outdated
Show resolved
Hide resolved
| @@ -581,7 +581,8 @@ public LocalExecutionPlan plan( | |||
| PartitionFunction partitionFunction; | |||
| Optional<SkewedPartitionRebalancer> skewedPartitionRebalancer = Optional.empty(); | |||
| int taskCount = getTaskCount(partitioningScheme); | |||
| if (checkCanScalePartitionsRemotely(taskContext.getSession(), taskCount, partitioningScheme.getPartitioning().getHandle(), nodePartitioningManager)) { | |||
| if (SystemSessionProperties.getRetryPolicy(taskContext.getSession()) != RetryPolicy.TASK | |||
There was a problem hiding this comment.
nit: what happens if I change policy from TASK to none via Session property?
There was a problem hiding this comment.
So NONE is no retries - and the we want to use skewedPartitionRebalancer
There was a problem hiding this comment.
I would guess using none on fte cluster has perf implications
| @@ -581,7 +581,8 @@ public LocalExecutionPlan plan( | |||
| PartitionFunction partitionFunction; | |||
There was a problem hiding this comment.
Is it somewhat testable in query plans or similar? Is it worth testing? I would assume there was regression due to scaling rebalancer in FTE
There was a problem hiding this comment.
I will check how much regression we get realistically - not tested - yet this code is not needed in FTE - so CPU cycles are wasted for sure.
losipiuk
force-pushed
the
lo/remove-skew-rebalancer-fte
branch
from
81b873a
to
f8a6adb
Compare
The rebalancing logic from SkewedPartitionRebalancer is already covered by partitions splitting/merging in HashDistributionSplitAssigner. Also the way remote SkewedPartitionRebalancer is configured does not match FTE behaviour as there is assumption that data will be consumed by N tasks where N is number of cluster nodes. This is not true for FTE.
losipiuk
force-pushed
the
lo/remove-skew-rebalancer-fte
branch
from
f8a6adb
to
5453fe9
Compare
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.
The rebalancing logic from SkewedPartitionRebalancer is already covered by partitions splitting/merging in HashDistributionSplitAssigner. Also the way remote SkewedPartitionRebalancer is configured does not match FTE behaviour as there is assumption that data will be consumed by N tasks where N is number of cluster nodes. This is not true for FTE.
Release notes
(x) This is not user-visible or is docs only, and no release notes are required.
( ) Release notes are required. Please propose a release note for me.
( ) Release notes are required, with the following suggested text: