beryllw opened a new issue, #3259:
URL: https://github.com/apache/fluss/issues/3259

   ### Search before asking
   
   - [x] I searched in the [issues](https://github.com/apache/fluss/issues) and 
found nothing similar.
   
   
   ### Description
   
   ## Problem
   The TieringCommitOperator currently runs with parallelism=1 and 
maxParallelism=1. All table commits are processed sequentially in a single 
thread. When a table's lake commit triggers manifest compaction (e.g., in 
Paimon), it can take a significant amount of time, during which all other 
tables' commits are blocked.This causes data freshness degradation for 
unrelated tables, potentially triggering alerts even though those tables have 
data ready to commit.
   
   ## Current Behavior
   The commit flow in TieringCommitOperator.processElement() is synchronous:
   1. Collect all bucket write results for a table
   2. Call LakeCommitter.toCommittable()
   3. Call LakeCommitter.commit() — this is the bottleneck, as it may trigger 
manifest compaction
   4. Call FlussTableLakeSnapshotCommitter.commit() to notify Fluss
   
   Since the operator is single-threaded with parallelism=1, step 3 for one 
table blocks steps 1-4 for all other tables.
   
   ### Willingness to contribute
   
   - [x] I'm willing to submit a PR!


-- 
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]

Reply via email to