DataStax OpsCenter for Cassandra - why is disk usage reported by storage capacity widget different than "Data Size" metric?
The Storage Capacity widget shows 38.96 TB used but a graph of the "Data Size" (data-load) metric (for the "cluster wide") shows 14.30 TB used. This cluster has two data centers and a RF of 3 for all tables.
Here is an example image of the Storage Capacity widget
Those are two different things and are not related to each other.
In DSE OpsCenter, the values displayed in the Storage Capacity widget on the cluster dashboard is an aggregate of used and free disk space of all the nodes as reported by the operating system.
Under the hood, OpsCenter uses native Linux utilities such as iostat and netstat which are part of the Sysstat package to collect operating system metrics. The Storage Capacity widget is based on the output reported by the Linux df utility.
On the other hand, the OpsCenter Data Size metric (data-load) is based on the Cassandra metric org.apache.cassandra.metrics.Storage.Load which represents the size of the data managed by the node. This is the same value reported in the Load column of the nodetool status output. Cheers!
UPDATE: See Monitoring sufficient disk space for restoring backups for additional info on the Storage Capacity widget.
