The GitHub Actions job "CI" on rocketmq-dashboard.git/fix/timeout-enrichment-sentinel has failed. Run started by GitHub user unbridled-41 (triggered by unbridled-41).
Head commit for run: 712b91799da9bc7111fa57be775eb095c4d7d3c3 / unbridled-41 <[email protected]> fix(group): report a cancelled group enrichment as unknown, not as zero online clients When the batch deadline cancels a group's enrichment task before it starts, the snapshot keeps its initial value: AtomicReference<GroupLiveSnapshot> snapshot = new AtomicReference<>(GroupLiveSnapshot.empty()); ... private static GroupLiveSnapshot empty() { return new GroupLiveSnapshot(List.of(), 0, false, ...); } `onlineInstances = 0` then reaches the API as a measured "this group has no online clients", while the same class maps a failed lookup to the documented sentinel (`connection.available() ? instances.size() : -1`, and ConsumerGroupVO says "-1 when the connection inventory is unavailable"). The consequences are user-visible: the consumer page renders a green 0 for the client column (the sentinel renders as 不可用) and the health verdict turns the zero into a critical "有堆积但无在线客户端" issue for a group whose inventory was never read. Initialize the snapshot with the sentinel (renamed `unknown()`), leaving every worker-written snapshot untouched. Tests (server/): mvn -o -B -ntp test -Dtest='RocketMQMetadataProviderTest' -> Tests run: 48, Failures: 0, Errors: 0. - timedOutGroupEnrichmentMustNotMutateReturnedGroupTest now asserts -1 instead of 0 at both points; it fails before this change (`expected: -1 but was: 0`). - mvn -o -B -ntp test -Dtest='MetadataServiceTest,ConsumerGroupControllerTest,RocketMQClientProviderTest' -> Tests run: 125, Failures: 0, Errors: 0; checkstyle passes. Report URL: https://github.com/apache/rocketmq-dashboard/actions/runs/37990990270 With regards, GitHub Actions via GitBox
