Aias00 opened a new issue, #2662: URL: https://github.com/apache/rocketmq-dashboard/issues/2662
## Problem `CollectorScheduler` submits every configured Studio instance to a fixed-size executor at once. The executor uses the same value for worker count and queue capacity, with `AbortPolicy`; rejected submissions are logged and discarded. With the default parallelism of 4, only four running plus four queued jobs are accepted. When more than eight instances exist, later instances in the stable repository order can be skipped on every collection pass, leaving them without metric snapshots or alert evaluation. ## Evidence - `server/src/main/java/org/apache/rocketmq/studio/cluster/metrics/CollectorScheduler.java:82` - `server/src/main/java/org/apache/rocketmq/studio/cluster/metrics/CollectorScheduler.java:102` - `server/src/main/java/org/apache/rocketmq/studio/cluster/metrics/CollectorScheduler.java:167` - `server/src/main/java/org/apache/rocketmq/studio/instance/MybatisPlusInstanceRepository.java:46` ## Expected behavior - Every configured instance gets a collection opportunity during a pass. - Concurrency remains bounded by `collectionParallelism`. - A global/pass deadline prevents an unbounded pass. - Saturation must not silently starve the same tail instances. - Regression coverage includes more than `2 * parallelism` instances. ## Related #2641 / PR #2642 handles collection lease renewal, not executor saturation or dropped jobs. -- 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]
