You will need to adjust the value of AMS heap size
When memstore are being forced to flush to make room in memory.keep flushing until we hit this mark. Defaults to 35% of heap. The value equals to the hbase.regionserver.global.memstore.upperLimit causes the minimum possible flushing to occur when updates are blocked to memstore limiting
Maximum Size of all memstores in the region server before new updates are blocked and flushes are forces. This defaults to 40% of the heap
So what is the current size of your Metrics Collector Heap Size? With the above setup with a cluster size of <20 nodes Set in the Advanced ams-env the value of Metrics Collector Heap Size = 1024 that should work.
Please can you use this as a reference to tune your AMS