AWS CloudFormation template for CloudWatch alarm: how to describe the alarm with metrics by ElasticSearch node, not by whole ElasticSearch cluster

Question

I have CloudFormation template for CloudWatch alarm which works with metrics from ElasticSearch cluster, here it is:

...
Resources:
  EsCPUUtilizationTooHighAlarm:
    Type: "AWS::CloudWatch::Alarm"
    Properties:
      AlarmName: "es-cpu-utilization-too-high"
      AlarmActions:
      - !Ref AlarmsSnsTopic
      AlarmDescription: "ES cluster CPU usage above 80%"
      ComparisonOperator: "GreaterThanThreshold"
      TreatMissingData: "missing"
      Dimensions:
      - Name: ClientId
        Value: !Ref "AWS::AccountId"
      - Name: DomainName
        Value: !Ref ElasticsearchCluster
      EvaluationPeriods: 1
      MetricName: "CPUUtilization"
      Namespace: "AWS/ES"
      Period: "300"
      Statistic: "Average"
      Threshold: "80"
...

This alarm tracks metrics from whole ElasticSearch cluster as you see, but I'd like to create few such alarms and track each ElasticSearch node separately. How can I describe such CloudWatch alarm with CloudFormation template?

Answer 1

The Managing Amazon Elasticsearch Service Domains documentation lists metrics that are available for each individual instance in the cluster, mostly around Thread Pools and JVM metrics.

There is a SysMemoryUtilization metric defined as The percentage of the instance's memory that is in use .