SEZ9 commented on PR #12081:
URL: https://github.com/apache/seatunnel/pull/12081#issuecomment-6620608374

   Thanks for the detailed answers against `ef99d22d1`, @Rangsh.
   
   **Field name (`writDataTimeoutMilliseconds`)** — agreed. Since the public 
config key is already `writeDataTimeoutMilliseconds` and the typo on the 
private field is pre-existing, renaming it here would be pure churn. Leaving it 
for a follow-up is fine with me. One note: F1 as filed was about the untimed 
`RequestFuture.get()` now blocking without bound if the WAL worker dies or the 
event is never dispatched, so a pointer on that would still help.
   
   **F2 / F4** — the shape you describe in `WALWorkHandler` (failure flag held 
for the handler lifetime at L57-L72, later APPEND events returning 
`done(false)` before the writer is touched at L95-L104) is what I was hoping 
for: the worker survives and the writer is not reused after a failed write. Two 
small things so I can sign these off:
   
   1. Your comment appears to have been cut off right after the L95-L104 
pointer. Could you re-post the rest of it?
   2. For F2 specifically, please confirm that the `executeResponse()` call is 
inside the guarded path, so a future already removed by the caller-side timeout 
cannot throw and take down the single disruptor worker. If that is covered in 
the truncated part or in comment 6558601142, a one-line pointer is enough.
   
   **Remaining items (F3, F5, F6, F7, F8)** — for each, could you either point 
me to where on `ef99d22d1` it is addressed, or state explicitly that it is 
deferred? In particular F5: I'd like to confirm how 
`HdfsWriterFlushSyncPathTest` gets Mockito on the classpath for the 
`imap-storage-file` module, since I did not see a `pom.xml` change in this PR.
   
   Once those are in, I'll do a final pass and move this forward.
   
   <!-- streview-comment:2254 -->


-- 
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