You are viewing a plain text version of this content. The canonical link for it is here.
Posted to github@datafusion.apache.org by "2010YOUY01 (via GitHub)" <gi...@apache.org> on 2025/04/01 06:55:17 UTC
[PR] feat: Add config max_temp_directory_size to limit max disk usage for spilling queries [datafusion]
2010YOUY01 opened a new pull request, #15520:
URL: https://github.com/apache/datafusion/pull/15520
## Which issue does this PR close?
<!--
We generally require a GitHub issue to be filed for all bug fixes and enhancements and this helps us generate change logs for our releases. You can link an issue to this PR using the GitHub syntax. For example `Closes #123` indicates that this PR will close issue #123.
-->
- Closes #15358
## Rationale for this change
See the rationale part of the first attempt PR https://github.com/apache/datafusion/pull/14975
The included changes and implementation are different than the above PR, it will be explained below.
<!--
Why are you proposing this change? If this is already explained clearly in the issue then this section is not needed.
Explaining clearly why changes are proposed helps reviewers understand your changes and offer better suggestions for fixes.
-->
## What changes are included in this PR?
<!--
There is no need to duplicate the description in the issue here but it is sometimes worth providing a summary of the individual changes in this PR.
-->
1. Add `max_temp_directory_size` field inside `DiskManager` to keep track of current total disk usage for temporary files, by default it's 100GB.
2. Update `RefCountedTempFile`:
- Included a reference to `DiskManager` that created it
- Added a utility function `update_disk_usage()` to update the global disk usage. After modifying the managed tempfile, the caller also has to call this function to do the update, to make sure when disk limit is exceeded an error will be thrown. (Currently `RefCountedTempFile` is only used for spill files inside DataFusion, so I think this additional interface is okay to add)
## Are these changes tested?
<!--
We typically require tests for all PRs in order to:
1. Prevent the code from being accidentally broken by subsequent changes
4. Serve as another way to document the expected behavior of the code
If tests are not included in your PR, please explain why (for example, are they covered by existing tests)?
-->
Yes, integration test is included for queries exceed/not-exceed the disk limit.
## Are there any user-facing changes?
<!--
If there are user-facing changes then we may require documentation to be updated before approving the PR.
-->
<!--
If there are any breaking changes to public APIs, please add the `api change` label.
-->
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] feat: Add config `max_temp_directory_size` to limit max disk usage for spilling queries [datafusion]
Posted by "2010YOUY01 (via GitHub)" <gi...@apache.org>.
2010YOUY01 commented on PR #15520:
URL: https://github.com/apache/datafusion/pull/15520#issuecomment-2774468597
@alamb Thank you. I filed https://github.com/apache/datafusion/issues/15553 for configuring this setting in `datafusion-cli`.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] feat: Add config `max_temp_directory_size` to limit max disk usage for spilling queries [datafusion]
Posted by "Copilot (via GitHub)" <gi...@apache.org>.
Copilot commented on code in PR #15520:
URL: https://github.com/apache/datafusion/pull/15520#discussion_r2022259546
##########
datafusion/execution/src/disk_manager.rs:
##########
@@ -164,6 +216,50 @@ impl RefCountedTempFile {
pub fn inner(&self) -> &NamedTempFile {
&self.tempfile
}
+
+ /// Updates the global disk usage counter after modifications to the underlying file.
+ ///
+ /// # Errors
+ /// - Returns an error if the global disk usage exceeds the configured limit.
+ pub fn update_disk_usage(&mut self) -> Result<()> {
+ // Get new file size from OS
+ let metadata = self.tempfile.as_file().metadata().unwrap();
Review Comment:
Using unwrap on file metadata could cause a panic if metadata retrieval fails; consider propagating the error instead of unwrapping.
```suggestion
let metadata = self.tempfile.as_file().metadata()?;
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] feat: Add config `max_temp_directory_size` to limit max disk usage for spilling queries [datafusion]
Posted by "alamb (via GitHub)" <gi...@apache.org>.
alamb merged PR #15520:
URL: https://github.com/apache/datafusion/pull/15520
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] feat: Add config `max_temp_directory_size` to limit max disk usage for spilling queries [datafusion]
Posted by "alamb (via GitHub)" <gi...@apache.org>.
alamb commented on PR #15520:
URL: https://github.com/apache/datafusion/pull/15520#issuecomment-2776693763
Thanks again @2010YOUY01
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] feat: Add config `max_temp_directory_size` to limit max disk usage for spilling queries [datafusion]
Posted by "alamb (via GitHub)" <gi...@apache.org>.
alamb commented on code in PR #15520:
URL: https://github.com/apache/datafusion/pull/15520#discussion_r2025546904
##########
datafusion/core/tests/memory_limit/mod.rs:
##########
@@ -524,6 +526,93 @@ async fn test_external_sort_zero_merge_reservation() {
assert!(spill_count > 0);
}
+// Tests for disk limit (`max_temp_directory_size` in `DiskManager`)
+// ------------------------------------------------------------------
+
+// Create a new `SessionContext` with speicified disk limit and memory pool limit
+async fn setup_context(
+ disk_limit: u64,
+ memory_pool_limit: usize,
+) -> Result<SessionContext> {
+ let disk_manager = DiskManager::try_new(DiskManagerConfig::NewOs)?;
+
+ let disk_manager = Arc::try_unwrap(disk_manager)
+ .expect("DiskManager should be a single instance")
+ .with_max_temp_directory_size(disk_limit)?;
+
+ let runtime = RuntimeEnvBuilder::new()
+ .with_memory_pool(Arc::new(FairSpillPool::new(memory_pool_limit)))
+ .build_arc()
+ .unwrap();
+
+ let runtime = Arc::new(RuntimeEnv {
+ memory_pool: runtime.memory_pool.clone(),
+ disk_manager: Arc::new(disk_manager),
+ cache_manager: runtime.cache_manager.clone(),
+ object_store_registry: runtime.object_store_registry.clone(),
+ });
+
+ let config = SessionConfig::new()
+ .with_sort_spill_reservation_bytes(10 * 1024 * 1024) // 10MB
+ .with_target_partitions(1);
+
+ Ok(SessionContext::new_with_config_rt(config, runtime))
+}
+
+/// If the spilled bytes exceed the disk limit, the query should fail
+/// (specified by `max_temp_directory_size` in `DiskManager`)
+#[tokio::test]
+async fn test_disk_spill_limit_reached() -> Result<()> {
+ let ctx = setup_context(10 * 1024 * 1024, 60 * 1024 * 1024).await?;
+
+ let df = ctx
+ .sql("select * from generate_series(1, 1000000000000) as t1(v1) order by v1")
+ .await
+ .unwrap();
+
+ let err = df.collect().await.unwrap_err();
+ assert_contains!(
+ err.to_string(),
+ "The used disk space during the spilling process has exceeded the allowable limit"
+ );
+
+ Ok(())
+}
+
+/// External query should succeed, if the spilled bytes is less than the disk limit
+/// Also verify that after the query is finished, all the disk usage accounted by
+/// tempfiles are cleaned up.
+#[tokio::test]
+async fn test_disk_spill_limit_not_reached() -> Result<()> {
+ let disk_spill_limit = 100 * 1024 * 1024; // 100MB
Review Comment:
do we really need to generate 100MB to test temporary file space? Could we perhaps lower this to something less resource intensive like 1MB (and reduce the argument to `generate_series`)?
##########
datafusion/physical-plan/src/spill/in_progress_spill_file.rs:
##########
@@ -70,6 +75,11 @@ impl InProgressSpillFile {
}
if let Some(writer) = &mut self.writer {
let (spilled_rows, spilled_bytes) = writer.write(batch)?;
+ if let Some(in_progress_file) = &mut self.in_progress_file {
+ in_progress_file.update_disk_usage()?;
Review Comment:
It is quite nice that this is encapsulated as part of `InProgressSpillFile`
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org
Re: [PR] feat: Add config `max_temp_directory_size` to limit max disk usage for spilling queries [datafusion]
Posted by "2010YOUY01 (via GitHub)" <gi...@apache.org>.
2010YOUY01 commented on code in PR #15520:
URL: https://github.com/apache/datafusion/pull/15520#discussion_r2026154931
##########
datafusion/core/tests/memory_limit/mod.rs:
##########
@@ -524,6 +526,93 @@ async fn test_external_sort_zero_merge_reservation() {
assert!(spill_count > 0);
}
+// Tests for disk limit (`max_temp_directory_size` in `DiskManager`)
+// ------------------------------------------------------------------
+
+// Create a new `SessionContext` with speicified disk limit and memory pool limit
+async fn setup_context(
+ disk_limit: u64,
+ memory_pool_limit: usize,
+) -> Result<SessionContext> {
+ let disk_manager = DiskManager::try_new(DiskManagerConfig::NewOs)?;
+
+ let disk_manager = Arc::try_unwrap(disk_manager)
+ .expect("DiskManager should be a single instance")
+ .with_max_temp_directory_size(disk_limit)?;
+
+ let runtime = RuntimeEnvBuilder::new()
+ .with_memory_pool(Arc::new(FairSpillPool::new(memory_pool_limit)))
+ .build_arc()
+ .unwrap();
+
+ let runtime = Arc::new(RuntimeEnv {
+ memory_pool: runtime.memory_pool.clone(),
+ disk_manager: Arc::new(disk_manager),
+ cache_manager: runtime.cache_manager.clone(),
+ object_store_registry: runtime.object_store_registry.clone(),
+ });
+
+ let config = SessionConfig::new()
+ .with_sort_spill_reservation_bytes(10 * 1024 * 1024) // 10MB
+ .with_target_partitions(1);
+
+ Ok(SessionContext::new_with_config_rt(config, runtime))
+}
+
+/// If the spilled bytes exceed the disk limit, the query should fail
+/// (specified by `max_temp_directory_size` in `DiskManager`)
+#[tokio::test]
+async fn test_disk_spill_limit_reached() -> Result<()> {
+ let ctx = setup_context(10 * 1024 * 1024, 60 * 1024 * 1024).await?;
+
+ let df = ctx
+ .sql("select * from generate_series(1, 1000000000000) as t1(v1) order by v1")
+ .await
+ .unwrap();
+
+ let err = df.collect().await.unwrap_err();
+ assert_contains!(
+ err.to_string(),
+ "The used disk space during the spilling process has exceeded the allowable limit"
+ );
+
+ Ok(())
+}
+
+/// External query should succeed, if the spilled bytes is less than the disk limit
+/// Also verify that after the query is finished, all the disk usage accounted by
+/// tempfiles are cleaned up.
+#[tokio::test]
+async fn test_disk_spill_limit_not_reached() -> Result<()> {
+ let disk_spill_limit = 100 * 1024 * 1024; // 100MB
Review Comment:
[5506637](https://github.com/apache/datafusion/pull/15520/commits/550663795933ae538a3f6864465f473765991df8) reduced the disk and memory usage of UT to < 1MB
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: github-unsubscribe@datafusion.apache.org
For additional commands, e-mail: github-help@datafusion.apache.org