Rangsh commented on PR #12081: URL: https://github.com/apache/seatunnel/pull/12081#issuecomment-6050833291
@SEZ9 Sorry about the truncation — reposting just the tail (items 5 and 6) as requested. 5. **`batchQueryExecuteFailsStatus` waiting the full `writDataTimeoutMilliseconds` per entry sequentially — fixed in this commit.** The loop now computes one shared `deadlineNanos` before iterating, and each entry waits only `Math.max(0, deadlineNanos - System.nanoTime())` instead of a fresh full timeout. A stuck WAL worker therefore stalls `storeAll`/`deleteAll` for at most ~1 × `writDataTimeoutMilliseconds` for the whole batch, not N × it (`IMapFileStorage.java` L354–377). Covered by `IMapFileStorageBatchDeadlineTest`, which asserts the total wait stays bounded by the single timeout when the first entry consumes the entire budget. 6. **Full `TimeoutException` stack trace at ERROR on every timed-out wait in `queryExecuteStatus` — downgraded, per your earlier review note.** Timeouts now log a single-line WARN carrying requestId / elapsed / limit, with the stack trace at DEBUG; ERROR + full stack is reserved for the unexpected `catch (Exception)` branch (`IMapFileStorage.java` L336–345). This is intentional, to avoid flooding ERROR logs for expected timeout paths. If you'd rather have ERROR + full stack on every timeout, happy to change it. Also, on 1a: understood that javadoc alone doesn't remove the hang risk. Since there are no production callers of the untimed `get()`, I've kept it unbounded for `Future` contract compliance, with the in-code justification in the method-level javadoc (`RequestFuture.java` L57–68). If you'd prefer a bounded wait (e.g. failing after `Long.MAX_VALUE`-style cap or a configurable ceiling), I'm happy to add one — just say which you'd like. -- 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]
