Add instanceId to health check log

[dil-ddecarvalhogomes]

Description of changes:

I need to create a log monitor alert for SSM Agents but since the instance Id is only present in the Log Stream, it's not possible to get the log stream and pack it as a dimension, making the alert useless.

Therefore, this is what I need, output the instanceId in the health check line:

Top couple lines are from the original AWS SSM Agent (no instance id) and bottom lines is after I installed my version.

Then, with the following AWS CDK v2 code, I was able to create the monitor that I needed:

  createSSMAgentHealthMetricFilter(
    context: InfraContext
  ): logs.MetricFilter {
    const pattern = logs.FilterPattern.spaceDelimited(
      "w1",
      "w2",
      "w3",
      "w4",
      "w5",
      "w6",
      "InstanceId",
      "w7",
      "w8",
      "..."
    )
      .whereString("w6", "=", "HealthCheck")
      .whereString("w8", "=", "reporting");

    const metricFilter = new logs.MetricFilter(
      this,
      "SSMAgentHealthLogFilter",
      {
        filterPattern: pattern,
        logGroup: context.ssmAgentLogGroup,
        metricNamespace: "SSM/Agent",
        metricName: "SSMAgentHealthLog",
        metricValue: "1",
        dimensions: {
          InstanceId: "$InstanceId",
        },
      }
    );
    
    return metricFilter;
  }

... instantiate alert in a different stack ...

  // create a CLoudWatch alert for the SSM Agent logs
  createSsmAgentHealthLogAlarm(
    context: ServiceContext,
    instance: ec2.Instance,
    instanceIndex: number
  ) {
    const metric = new cloudwatch.Metric({
      namespace: "SSM/Agent",
      metricName: "SSMAgentHealthLog",
      statistic: "Sum",
      dimensionsMap: {
        InstanceId: instance.instanceId,
      },
    });

    new cloudwatch.Alarm(this, `SSMAgentErrorAlarm${instanceIndex}`, {
      metric: metric,
      threshold: 1,
      evaluationPeriods: 5,
      comparisonOperator: cloudwatch.ComparisonOperator.LESS_THAN_THRESHOLD,
      alarmName: "SSMAgentHealthAlarm[" + instance.instanceId + "]",
      alarmDescription:
        "SSM Agent not reporting health checks on " + instance.instanceId,
      actionsEnabled: true,
      treatMissingData: cloudwatch.TreatMissingData.BREACHING,
    });
  }

Now, a reasonable alternative would be to install the CloudWatch agent and configure a process monitor, which requires a PowerShell script. This is a LOT of hops to get the SSM agent monitored.

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.