SEZ9 commented on PR #12081:
URL: https://github.com/apache/seatunnel/pull/12081#issuecomment-6994414973

   Thanks for the update. Following up on the earlier review points only:
   
   **Addressed in the latest recheck**
   - **`RequestFuture` overload contracts (compatibility and docs):** both 
`get()` overloads are now documented, so the method-level contract gap is 
closed.
   - **Worker dying on `executeResponse()`:** the cache lookup and response 
callback are now protected, which covers the timeout-plus-remove race.
   - **Writer reuse after a write failure:** the first append failure marks the 
handler fail-closed and subsequent APPENDs fail, so the torn-record concern is 
resolved.
   - **Sequential per-entry wait in the batch path:** there is now a single 
post-publication batch deadline, so there is no more N × timeout stacking.
   - **ERROR-level timeout stack traces:** timeouts now log a WARN message, 
with the stack at DEBUG.
   
   **Still open: two things I could not confirm from the thread**
   1. **Unbounded `get()`:** the untimed `get()` is documented, but is any 
production caller still using it? If the WAL worker dies via an `Error`, or the 
event is never dispatched, does that caller still block forever, or do all 
callers now go through the timed overload?
   2. **Mockito in `HdfsWriterFlushSyncPathTest`:** how does the Mockito test 
dependency get onto the imap-storage-file module's test classpath? If it is 
inherited, please point to where. Otherwise the module needs the test-scoped 
dependency added.
   
   Once those two are confirmed, the earlier points are all settled from my 
side.
   
   <!-- streview-comment:2443 -->


-- 
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