wForget opened a new pull request, #12595:
URL: https://github.com/apache/gluten/pull/12595

   ## What changes are proposed in this pull request?
   
   Fix the task attempt ID used to generate Uniffle block IDs in the Velox 
columnar shuffle path.
   
   Previously, `UniffleShuffleManager#getWriter` passed Spark's globally 
increasing
   `TaskContext.taskAttemptId()` to `VeloxUniffleColumnarShuffleWriter`. This 
value may exceed the
   number of bits reserved for task attempt IDs in Uniffle's `BlockIdLayout`.
   
   This patch derives the task attempt ID using Uniffle's standard 
implementation:
   
   ```java
   getTaskAttemptIdForBlockId(context.partitionId(), context.attemptNumber())
   ```
   
   The derived stage-scoped ID is passed to the columnar shuffle writer, 
matching the behavior of
   Uniffle's standard `RssShuffleManager`. The string-form task ID remains 
unchanged.
   
   Fixes #12594
   
   ## How was this patch tested?
   
   existing uniffle integration tests.
   
   ## Was this patch authored or co-authored using generative AI tooling?
   
   Generated-by: OpenAI Codex (GPT-5)


-- 
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: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to