The Heatmaps view gives you a graphic representation of the overall utilization of your cluster using simple color coding.
Each host in the cluster is represented by a block. To see more information on a specific host, hover over the block you are interested in, and a popup with key host data appears. The color of the blocks represents usage in an appropriate unit based on a selectable set of metrics. If the data necessary to determine state are not all available, the block is marked as having Invalid Data. Changing the default maximum values for the heatmap lets you fine tune the representation. Use the Select Metric dropdown to select the metric type.
Currently the following metrics are supported:
Host/Disk Space Used % Uses disk.disk_free and disk.disk_total
Host/Memory Used %: Uses memory.mem_free and memory.mem_total
Host/CPU Wait I/O %: Uses cpu.cpu_wio
HDFS/Bytes Read: Uses dfs.datanode.bytes_read
HDFS/Bytes Written: Uses dfs.datanode.bytes_written
HDFS/Garbage Collection Time: Uses jvm.gcTimeMillis
HDFS/JVM Heap MemoryUsed: Uses jvm.memHeapUsedM
YARN/Garbage Collection Time: Uses jvm.gcTimeMillis
YARN / JVM Heap Memory Used: Uses jvm.memHeapUsedM
YARN / Memory used %: Uses UsedMemoryMB and AvailableMemoryMB
HBase/RegionServer read request count: Uses hbase.regionserver.readRequestsCount
HBase/RegionServer write request count: Uses hbase.regionserver.writeRequestsCount
HBase/RegionServer compaction queue size: Uses hbase.regionserver.compactionQueueSize
HBase/RegionServer regions: Uses hbase.regionserver.regions
HBase/RegionServer memstore sizes: Uses hbase.regionserver.memstoreSizeMB
