SEZ9 commented on PR #12081:
URL: https://github.com/apache/seatunnel/pull/12081#issuecomment-6841935179

   @Rangsh thanks for the pointers on `ef99d22d1` and `43e18b1e6`. I'll check 
the diff to confirm that the untimed `get()` delegates to the timed variant 
using the configured `writDataTimeoutMilliseconds`, that both `get` variants 
carry method-level Javadoc for the `TimeoutException` contract, and that the 
duplicate `mockito-junit-jupiter` dependency is gone from 
`imap-storage-file/pom.xml` and provided in test scope by the parent POM as 
@nzw921rx noted. If that all checks out, the unbounded-wait, docs, and Mockito 
dependency points are covered from my side.
   
   A few items from the earlier review still need clarification:
   
   1. **WALWorkHandler survivability** — you mention 
`WALWorkHandlerSurvivabilityTest` passes locally. Does it cover the case where 
`executeResponse()` throws (e.g. the future was already removed by the caller 
after a timeout) and the single disruptor worker keeps processing subsequent 
events?
   2. **Writer reuse after a failed write** — what happens to the writer after 
an arbitrary write failure in `WALWorkHandler`? If it is reused as-is, how is a 
torn record mid-file avoided (or recovered on replay)?
   3. **Sequential wait in `batchQueryExecuteFailsStatus`** — with each entry 
waiting up to the full write timeout, a stuck worker can block 
`storeAll`/`deleteAll` for N × timeout. Is there a plan to bound the total wait 
(e.g. a shared deadline across the batch), or is this intentional?
   4. **ERROR-level stack traces on timeout in `queryExecuteStatus`** — since a 
timed-out wait is now an expected, handled path, could this be logged at WARN 
without the full stack trace (or with the stack only at DEBUG)?
   
   Happy to re-review once these are addressed or clarified.
   
   <!-- streview-comment:2307 -->


-- 
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