I have worked on tests related to “interrupt”. Not sure about SPARK-50888:

My findings:
1. These test failures only occur in the GitHub CI.
2. The failure is due to the thread pool we created in CI having only two 
threads, while our tests require three concurrent threads to run.

To workaround this issue we could split these tests into smaller ones that 
require only two threads.
I’ll see if there’s a volunteer. Otherwise I could take a stab at fixing them.

Cheers,
Paddy

On 2025/01/19 07:52:00 Dongjoon Hyun wrote:
> Hi, All.
> 
> This is a kind of head-up as a part of Apache Spark 4.0.0 preparation.
> 
> https://issues.apache.org/jira/browse/SPARK-44111
> (Prepare Apache Spark 4.0.0)
> 
> It would be great if we are able to fix long-standing `Spark Connect` test
> flakiness
> together during the QA period (2025-02-01 ~) in order to make sure that we
> didn't
> miss any bugs here.
> 
> SPARK-48139: Re-enable `SparkSessionE2ESuite.interrupt tag`
> SPARK-50205: Re-enable
> `SparkSessionJobTaggingAndCancellationSuite.Cancellation APIs in
> SparkSession are isolated`
> SPARK-50748: Fix a flaky test: `SparkSessionE2ESuite.interrupt all -
> background queries, foreground interrupt`
> SPARK-50888: Fix Flaky Test: `SparkConnectServiceSuite.SPARK-44776:
> LocalTableScanExe`
> SPARK-50889: Fix Flaky Test: `SparkSessionE2ESuite.interrupt operation`
> (Hang)
> 
> Since many people consider the `Spark Connect` module important,
> some of JIRA issues are marked as `Blocker` priority for 4.0.0 currently.
> So, I hope we can make them re-enabled and stable to be safe.
> 
> However, we can also simply remove `Blocker` status before RC1 (2025-02-15)
> and ignore all of those test cases if `Spark Connect` module experts are
> sure that
> there are no `Spark Connect` functional issues behind these.
> In this case, the above issues will be re-scoped to Spark 4.1.0 as `Minor`
> priorities.
> 
> Sincerely,
> Dongjoon.
>  

Cheers,
Paddy


---------------------------------------------------------------------
To unsubscribe e-mail: dev-unsubscr...@spark.apache.org

Reply via email to