123123213weqw opened a new pull request, #2525: URL: https://github.com/apache/rocketmq-dashboard/pull/2525
## What is the purpose of the change `fillCountsInParallel` submitted one task per instance and then waited on each future with its own 3s `get()`. A batch of hung vendor calls therefore cost `3s × instance-count` of response latency instead of a single 3s wait. ## Brief changelog - submit the whole batch through `ExecutorService.invokeAll` with one shared `COUNT_TIMEOUT_SECONDS` deadline - tasks that miss the deadline are cancelled (interrupted) by `invokeAll`; their rows are marked counts-unavailable like any other failure - a test with three hung providers asserts the batch is bounded to well under one timeout per instance and that every row degrades to counts-unavailable ## How was this patch verified - server: `InstanceServiceTest` 69/69, including the new single-deadline test; full `mvn test` ran 1527 tests with only the 7 pre-existing environment failures in the CLI agent tests (missing `sh` binary on a Windows machine, identical on the clean base) Fixes #2495 -- 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]
