Discrepancy in memory consumption data between Cluster Utilization dashboard and Grafana

Issue

• There is a discrepancy in information about memory consumption between OpenShift > Home > Overview > Cluster Utilization dashborad in RHOCP and (custom) Grafana dashboard.
  • The full Grafana rule is the following:

  "targets": [
    {
      "expr": "sum (container_memory_working_set_bytes{name!=\"\",node=~\"^$Node$\"}) by (pod_name)",
      "format": "time_series",
      "interval": "1s",
      "intervalFactor": 1,
      "legendFormat": "{{pod_name}}",
      "metric": "container_memory_usage:sort_desc",
      "refId": "A",
      "step": 10,
      "datasource": {
        "uid": "$datasource"
      }
    }
  ]

• The RHOCP dashboard displays 1.9TB memory usage while Grafana dashboard reveals 2.8TB.