Cancellation benchmark #14818
Merged
Cancellation benchmark #14818
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
carols10cents
force-pushed
the
cancellation-test-case
branch
from
187e64f to
ca6074f
Compare
Contributor
Author
|
Let's see if this works too.... /benchmark |
Connects to apache#14036. This benchmark loads multiple files into an in-memory object store, starts a datafusion query in a new tokio runtime, lets the query run for an amount of time, cancels the query, and measures how long it takes to drop the tokio runtime. This demonstrates datafusion is likely not yielding often enough to allow for timely query cancellation and freeing up of all resources.
carols10cents
force-pushed
the
cancellation-test-case
branch
from
ca6074f to
a768d25
Compare
This was referenced Feb 23, 2025
Closed
Contributor
It doesn't sadly -- we removed this due to potential security concerns |
Contributor
Author
Ok-- I'll remove the documentation about it in my benchmark docs PR. |
alamb
approved these changes
Feb 26, 2025
Contributor
There was a problem hiding this comment.
Thanks @carols10cents
I tested this PR with
./benchmarks/bench.sh run cancellationAnd it looked like this (👍 )
benchmarks/data/cancellation -o /Users/andrewlamb/Software/datafusion/benchmarks/results/cancellation-test-case/cancellation.json`
No data files found, generating (this will take a bit)
Generating file 1 of 7
Generating file 2 of 7
Generating file 3 of 7
Generating file 4 of 7
Generating file 5 of 7
Generating file 6 of 7
Generating file 7 of 7
Done generating files
Using 7 files now on disk
Starting to load data into in-memory object store
Done loading data into in-memory object store
in main, sleeping
Starting spawned
Creating logical plan...
Creating physical plan...
Executing physical plan...
Getting results...
cancelling thread
done dropping runtime in 32.72425ms
Iteration 0 cancelled in 32.724250000000005 ms
in main, sleeping
...
I made a small comment suggestion, but I think we can do it as a follow on PR
|
|
||
| /// Test performance of cancelling queries | ||
| /// | ||
| /// The queries are executed on a synthetic dataset generated during |
Contributor
There was a problem hiding this comment.
Suggested change
| /// The queries are executed on a synthetic dataset generated during | |
| /// Queries in DataFusion should stop executing "quickly" after they are | |
| /// cancelled (the output stream is dropped). | |
| /// | |
| /// The queries are executed on a synthetic dataset generated during |
Contributor
|
Thanks again @carols10cents |
This was referenced Mar 6, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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.
Which issue does this PR close?
Rationale for this change
The behavior observed in #14036 was hard to reproduce and quantify; having a benchmark makes that easier!
What changes are included in this PR?
This benchmark loads multiple files into an in-memory object store, starts a datafusion query in a new tokio runtime, lets the query run for an amount of time, cancels the query, and measures how long it takes to drop the tokio runtime.
This demonstrates datafusion is likely not yielding often enough to allow for timely query cancellation and freeing up of all resources.
Are these changes tested?
This PR is only tests :)
Are there any user-facing changes?
Nope!