SEZ9 commented on PR #12161:
URL: https://github.com/apache/seatunnel/pull/12161#issuecomment-5594979793
Thanks @DanielLeens — appreciated the independent re-verification against
`88cc6b851818`. Your reading matches mine: with the send and the bounded wait
living in the same per-member lambda of a sequential stream, member N+1 is
never dispatched before member N resolves or times out, so the endpoint is
still bounded by 3s × member count rather than a single 3s ceiling. Agreed on
the fix direction too: fire all `sendOperationToMemberNode` calls first,
collect the futures, then await them against one shared deadline (or
`CompletableFuture.allOf(...).get(3, SECONDS)` if they get adapted). That also
folds the F4 interrupt handling into one place, which is the right shape.
To be explicit about what still needs to land in the next revision:
- **F2 (blocking):** dispatch-all-then-await against a shared deadline as
described above.
- **F4 (blocking):** on `InterruptedException`, stop iterating rather than
continuing and returning 200 with silent partial data; log it once.
- **F1 (blocking):** don't render a timed-out member as an anonymous `{}` —
carry at least the member address plus a timeout/error marker so it is
distinguishable from an empty node, and make the timeout configurable instead
of hardcoded.
- **F6:** extend `SeaTunnelHealthMonitorTest` to cover the
`renderLoad`/`renderOperationService` sites and avoid mutating the JVM-global
default locale.
- **F3:** document the timeout behaviour of `/system-monitoring-information`
in the REST API docs (en/zh, v1/v2).
- **F5:** drop the duplicate `TimeUnit` import and run Spotless on the test
source so CI isn't blocked.
- **F7:** either drop `invocationFuture.cancel(true)` or add a comment
making clear it only abandons the local wait and does not stop the remote
operation.
Once a new commit is up I'll re-check F2/F4 together on the reworked loop;
ping me when it's ready.
<!-- streview-comment:913 -->
--
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]
