aglinxinyuan opened a new issue, #8175:
URL: https://github.com/apache/texera/issues/8175

   ### Task Summary
   
   Two fault-tolerance classes in the replay-log path have **no spec of their 
own**:
   
   - **`AsyncReplayLogWriter`** (38/44 fully covered) — the background thread 
that batches log records and flushes them. Its terminated-writer guards, its 
flush-interval sleep and its drain-on-terminate path are unexercised.
   - **`ReplayLogManagerImpl`** (30/34) — `markAsReplayDestination` is never 
called by any test.
   
   `CongestionControl` (30/34) additionally never reaches its `ssThreshold` 
clamp, which is only taken after five consecutive ack timeouts.
   
   **A hazard worth stating before anyone writes these tests.** 
`AsyncReplayLogWriter.terminate()` does an untimed `CompletableFuture.get()` 
that is only completed at the end of `run()`. Calling `terminate()` on a writer 
that was never `start()`ed **blocks forever**, and since amber's suites are 
strictly serial in one JVM (`Tags.limit(Tags.Test, 1)`), that hangs the whole 
module build rather than failing one test. Every test must `start()` before 
`terminate()`, with `terminate()` in a `finally` — the class extends `Thread`, 
so a missed one also leaks a live thread into the shared JVM.
   
   ### Task Type
   
   - [x] Testing / QA
   
   ### Was this issue authored using generative AI tooling?
   
   Generated-by: Claude Code (Opus 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]

Reply via email to