[SPARK-33173][CORE][TESTS][FOLLOWUP] Use local[2] and AtomicInteger
#30823
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
|
cc @fsamuel-bs @dongjoon-hyun @HyukjinKwon @viirya @mridulm Please take a look, thanks! |
srowen
approved these changes
Dec 17, 2020
|
Kubernetes integration test starting |
|
Kubernetes integration test status success |
|
Test build #132956 has finished for PR 30823 at commit
|
dongjoon-hyun
changed the title
local[2] and AtomicInteger
dongjoon-hyun
pushed a commit
that referenced
this pull request
Dec 17, 2020
### What changes were proposed in this pull request? Use `local[2]` to let tasks launch at the same time. And change counters (`numOnTaskXXX`) to `AtomicInteger` type to ensure thread safe. ### Why are the changes needed? The test is still flaky after the fix #30072. See: https://github.com/apache/spark/pull/30728/checks?check_run_id=1557987642 And it's easy to reproduce if you test it multiple times (e.g. 100) locally. The test sets up a stage with 2 tasks to run on an executor with 1 core. So these 2 tasks have to be launched one by one. The task-2 will be launched after task-1 fails. However, since we don't retry failed task in local mode (MAX_LOCAL_TASK_FAILURES = 1), the stage will abort right away after task-1 fail and cancels the running task-2 at the same time. There's a chance that task-2 gets canceled before calling `PluginContainer.onTaskStart`, which leads to the test failure. ### Does this PR introduce _any_ user-facing change? No ### How was this patch tested? Tested manually after the fix and the test is no longer flaky. Closes #30823 from Ngone51/debug-flaky-spark-33088. Authored-by: yi.wu <yi.wu@databricks.com> Signed-off-by: Dongjoon Hyun <dongjoon@apache.org> (cherry picked from commit 15616f4) Signed-off-by: Dongjoon Hyun <dongjoon@apache.org>
|
Since this test is added at SPARK-33088, merged to master/3.1. |
|
Late +1 |
|
Nice, thanks @Ngone51. |
|
thanks all! |
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.
What changes were proposed in this pull request?
Use
local[2]to let tasks launch at the same time. And change counters (numOnTaskXXX) toAtomicIntegertype to ensure thread safe.Why are the changes needed?
This test is added at SPARK-33088 and still flaky after the fix #30072. See: https://github.com/apache/spark/pull/30728/checks?check_run_id=1557987642
And it's easy to reproduce if you test it multiple times (e.g. 100) locally.
The test sets up a stage with 2 tasks to run on an executor with 1 core. So these 2 tasks have to be launched one by one.
The task-2 will be launched after task-1 fails. However, since we don't retry failed task in local mode (MAX_LOCAL_TASK_FAILURES = 1), the stage will abort right away after task-1 fail and cancels the running task-2 at the same time. There's a chance that task-2 gets canceled before calling
PluginContainer.onTaskStart, which leads to the test failure.Does this PR introduce any user-facing change?
No
How was this patch tested?
Tested manually after the fix and the test is no longer flaky.