Example for collecting memory usage using CloudWatch agent

[tschaub]

I've been struggling to configure an Elastic Beanstalk application to generate metrics on memory usage. In this repo, I've found the autoscaling-memory-utilization.config, and that seems to have the required parts. I've been having trouble getting it to work ("ERROR: Cannot obtain instance id from EC2 meta-data.") and found a doc on monitoring scripts that say they are deprecated. It sounds like the alternative is to "use the CloudWatch agent to collect metrics and logs."

Is there a complete example of how to use the CloudWatch agent to monitor memory use on Elastic Beanstalk instances? I'm not clear on the difference between CloudWatch agent and CloudWatch Logs agent, but I see that the latter is on the path to deprecation.

[tschaub]

Maybe I should add that on my EC2 instances, I already see configuration files in /opt/aws/amazon-cloudwatch-agent/etc/. So I guess I'm looking for a way to extend whatever configuration is there with memory metrics collection configuration.

[tschaub]

It looks like it is possible to append a CloudWatch agent configuration with amazon-cloudwatch-agent-ctl. I was able to successfully deliver memory use related metrics to CloudWatch by adding a .platform/confighooks/postdeploy/configure-cloudwatch-agent.sh script that looks like this:

#!/usr/bin/env bash
set -e

cat >/tmp/cloudwatch-agent-custom-config.json <<EOL
{
  "metrics": {
    "metrics_collected": {
      "mem": {
        "measurement": [
          "mem_used_percent"
        ],
        "metrics_collection_interval": 1
      }
    }
  }
}
EOL

amazon-cloudwatch-agent-ctl \
  -a append-config \
  -m ec2 \
  -c file:/tmp/cloudwatch-agent-custom-config.json \
  -s