Add additional EFA metrics

[the-mann]

Description

Adding the following EFA metrics:

• unresponsive_remote_events
• impaired_remote_conn_events
• retrans_timeout_events
• retrans_pkts
• retrans_bytes

for node, container, and pod level.

Testing

Manual testing:

apiVersion: eksctl.io/v1alpha5
kind: ClusterConfig

metadata:
  name: test-cwa-efa-cluster
  region: us-west-2
  version: "1.33"

iam:
  withOIDC: true

availabilityZones: ["us-west-2a", "us-west-2c"]

managedNodeGroups:
  - name: my-efa-ng-2
    instanceType: c6in.32xlarge
    minSize: 1
    desiredCapacity: 1
    maxSize: 1
    availabilityZones: ["us-west-2c"]
    volumeSize: 300
    privateNetworking: true
    efaEnabled: true

save that to cluster.yaml, and run

eksctl create cluster -f cluster.yaml

install the cloudwatch observability EKS addon

https://docs.aws.amazon.com/AmazonCloudWatch/latest/monitoring/Container-Insights-setup-EKS-addon.html

Then, deploy the following:

apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: my-training-job-2
  namespace: default
  labels:
    app: my-training-job-2
spec:
  selector:
    matchLabels:
      app: my-training-job-2
  template:
    metadata:
      labels:
        app: my-training-job-2
    spec:
      containers:
        - name: efa-device-holder
          image: busybox:latest
          command: ["/bin/sh", "-c", "sleep infinity"]
          resources:
            limits:
              vpc.amazonaws.com/efa: 1  # Request EFA device
            requests:
              vpc.amazonaws.com/efa: 1

finally, build a dev version of the cloudwatch agent with this branch + aws/amazon-cloudwatch-agent@3cc6b58

and check that the metrics exist:

[github-actions]

This PR was marked stale due to lack of activity. It will be closed in 14 days.

[duhminick]

typo in const name