It adds description to the `Ingress Percentile Response Times and Transfer Rates` view, so that user knows that this latency data is independant of dashboard time range.
This PR also adds two new panel about the latency, where one shows latency as a timeseries graph and other shows heatmap of the latency distribution.
nginx_ingress_controller_nginx_process_connections returns four elements
for each pod, one for each "state" (active, reading, waiting, writing).
The value of the element with state "active" is the sum of the other
three elements of other states:
active = reading + waiting + writing
So sum() returns a value that is 2x of the actual amount of connections.
To fix this we simply select elements with state "active".
