Metricbeat - system.memory.actual.used.pct


When i try to find the root cause of a crash using metricbeat field system.memory.actual.used.pct, my dashboard appears to change the values when i filter by for the machine that crashed. Before I filter by the host, the values are much lower, and the higher utilisations are coming from a different host. See attached gif.

Any ideas much appreciated.


It seems like the heatmap vis is not configured correctly. It's using the "percentage mode" which has a confusing name because what it's doing is that it's taking the current range of the data and treating that as 0-100%. In this mode there is at least one cell in the heatmap colored red no matter the value (e.g. if all values are between 0 and 0.1, then 0.1 is "100%" for the chart).

As system.memory.actual.used.pct is already a percentage, this behavior is pretty confusing, the chart should simply configure stable ranges from 0 to 1:


excellent. I swapped it for the custom ranges and it appears to be performing better.

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.