Please be aware, the engine metrics refer to the cpu/mem of each engine, not the jmeter metrics controller collects.
When multiple controller are pointing to the same k8s cluster, we can have redundant engine metrics. In the past, the engine metrics are managed by a background goroutine. But if the collection is not triggered by this particular controller, the local store will not have the meta data of the engines hence when they are going to be deleted, they cannot be deleted. With this PR, we move the engine metrics from local thread to purge. So when the engines are purged, the metrics are also purged.
Please be aware, the engine metrics refer to the cpu/mem of each engine, not the jmeter metrics controller collects.
When multiple controller are pointing to the same k8s cluster, we can have redundant engine metrics. In the past, the engine metrics are managed by a background goroutine. But if the collection is not triggered by this particular controller, the local store will not have the meta data of the engines hence when they are going to be deleted, they cannot be deleted. With this PR, we move the engine metrics from local thread to purge. So when the engines are purged, the metrics are also purged.