We see a case that we don't understand. In this print screen, you can see the Total Time metric value of 1020 ms and 1105 ms for min and max respectively. The problem is that the metric has different min and max values with the count of 1. This doesn't make sense, It should be the same value for min and max as there is only one count.
Release : 20.2
Component : APM Agents
This is working as designed:
- ASM Agent is polling data every 5 minutes, but the data harvesting and reporting is every 7.5 second. So most data points showing on the UI has count as 0, meaning no new data came in, so same old data is reported again. This is EPAgent default behavior for periods with no new data coming in.
- So depends on the timing, the harvest might be covering partial non-reporting time as well, for which the previous value is carried over as min/max.
Agent reports every 7.5 seconds, while EM harvest every 15 seconds. So if EM aggregates 2 data points from Agent, the result could be different min/max from the 2 data points.
- 1:23:20 [count=0, min/max=1105]
- 1:23:38 [count=0, min/max=1105]
- 1:28:45 [count=1, min/max=1020]
- 1:28:53 [count=0, min/max=1020]
- 1:29:00 [count=0, min/max=1020]
The EM data point of 1:28:45 is an aggregation of Agent data point of 1:23:38 and 1:28:45, that's how we get [count=1, min=1020, max=1105]