cluster:usage:consumption:rhods:cpu:seconds:rate1h:

sum(rate(container_cpu_usage_seconds_total

[1h]))

This should allow us to get a 1:1 match for what we expect to see in swatch w/ some simple adjustments (using a step of 3600 should show us the values swatch is ingesting; using a step of 24h and multiplying * 24 should show us the daily total swatch should be getting).