JVM pool metrics in Elasticsearch metricbeat

[renekalff]

Describe the enhancement: The JVM memory pool usage metrics for young, old and survivor are collected by the Elasticsearch module.

Describe a specific use case for the enhancement or feature: A key indicator of node memory pressure is the fill rate of the old memory pool. This is calculated by dividing the old pool used memory by the maximum old pool size. These metrics used to be collected by the Elasticsearch module and are still in the exported fields but aren't collected anymore.

The data from /_nodes/stats endpoint (only relevant fields for this feature request are shown):

"jvm": {
  "mem": {
    "pools": {
      "young": {
        "used_in_bytes": 1325400064,
        "max_in_bytes": 0,
        "peak_used_in_bytes": 7944011776,
        "peak_max_in_bytes": 0
      },
      "old": {
        "used_in_bytes": 5001717392,
        "max_in_bytes": 13421772800,
        "peak_used_in_bytes": 10183535520,
        "peak_max_in_bytes": 13421772800
      },
      "survivor": {
        "used_in_bytes": 213909504,
        "max_in_bytes": 0,
        "peak_used_in_bytes": 922746880,
        "peak_max_in_bytes": 0
      }
    }
  }
}

More information: https://www.elastic.co/blog/managing-and-troubleshooting-elasticsearch-memory https://www.elastic.co/blog/found-understanding-memory-pressure-indicator

[botelastic[bot]]

This issue doesn't have a Team:<team> label.