[Stack Monitoring] Stop displaying heap limit in relation to RSS for Kibana Memory Size

[simianhacker]

On the Memory Size chart for Kibana, we display "Heap Limit" in relationship to "Memory Usage" (which is RSS). The problem with this is that the heap is a subset of RSS which will also lead to scenarios where the "Memory Usage" will go above "Heap Limit". We should probably have 2 charts, one with RSS and one with heap usage (in relation to limits).

Current Fields we are using:

• Memory Usage - kibana_stats.process.memory.resident_set_size_in_bytes
• Heap Limit - kibana_stats.process.memory.heap.size_limit

CC: @joshdover

[elasticmachine]

Pinging @elastic/logs-metrics-ui (Team:logs-metrics-ui)

[ravikesarwani]

If I understand correctly the SDH in 1483 didn't had the heap size limit set and hence was giving 1400 MB as the default limit value. In my instance in Elastic Cloud I see that we do set the heap size limit (I see heap.size_limit as 851,443,710 for a 1GB kibana instance).

Looks like 2 things I am thinking we change/improve here:

• Add a line chart in the same visualization for Heap total (kibana_stats.process.memory.heap.total_in_bytes). This helps to compare it against heap size limit.
• Modify the descriptions to make things more clear
  • Heap size limit (kibana_stats.process.memory.heap.size_limit): Max. old space size allocated
  • Heap total (kibana_stats.process.memory.heap.total_in_bytes): Total size of the allocated heap
  • Memory usage (kibana_stats.process.memory.resident_set_size_in_bytes): Resident set size that includes code segment, stack and heap

I am not too sure if there’s a real need to move Resident set size (RSS) in a separate graph since it shows additional memory being used and probably need to be looked in the context of allocated heap to know how much extra (in code segment and stack) is being used.

Thoughts?

[elasticmachine]

Pinging @elastic/infra-monitoring-ui (Team:Monitoring)