AWS AutoScalingGroup Activity Exporter

A Prometheus exporter for AWS AutoScalingGroup Actvities

Helm Installation

AutoScalingGroup Actvitiy Exporter Helm Chart

Building and running

Source

Installing from source requries python >= 3.7 (install platform dependent)

Install poetry if needed with:

curl -sSL https://raw.githubusercontent.com/python-poetry/poetry/master/get-poetry.py | python -

poetry shell
poetry install

cd aws_autoscalinggroup_activity_exporter

# See Configuration section for details
cp conf/example-config.yaml conf/config.yaml

aws-autoscaling-group-activity-exporter --region <region-name> --port <port>

OR

aws-autoscaling-group-activity-exporter --region us-west-2 --port 8080

View metrics with curl localhost:8080/metrics

Docker

docker build -t aws-autoscalinggroup-activity-exporter:latest .

# no session token
docker run -e AWS_ACCESS_KEY_ID=$AWS_ACCESS_KEY_ID -e AWS_SECRET_ACCESS_KEY=$AWS_SECRET_ACCESS_KEY -it -d -p 8080:8080 aws-autoscalinggroup-activity-exporter:latest

OR

# using a seesion token
docker run -e AWS_ACCESS_KEY_ID=$AWS_ACCESS_KEY_ID -e AWS_SECRET_ACCESS_KEY=$AWS_SECRET_ACCESS_KEY -e AWS_SESSION_TOKEN=$AWS_SESSION_TOKEN -it -d -p 8080:8080 aws-autoscalinggroup-activity-exporter:latest

docker logs <container> -f

View metrics with curl localhost:8080/metrics

Credentials and permissions

Currently AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY can be used to run locally or in docker. Alternatively, if running on an EC2 instance or in EKS, the instance role can be used.

2 IAM Permissions are required and shown in the following example policy:

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "VisualEditor0",
            "Effect": "Allow",
            "Action": [
                "autoscaling:DescribeScalingActivities",
                "autoscaling:DescribeAutoScalingGroups"
            ],
            "Resource": "*"
        }
    ]
}

Configuration

The configuration is in YAML and must be stored in conf/config.yaml. An example can be found in conf/example-config.yaml

If tags are not specified, then all AutoScalingGroups in the region will gathered and metrics published for.

Custom causes can be added as well. Examples are given, but if there is a cause pattern you wish to capture, then you just need to add it with the cause label you wish to assign it.

Using the following example, if a cause with pattern:

'Spot Instance interruption notice' is detected, then it will publish a metric like so:

aws_autoscalinggroup_activity{instance_id="i-06b1e395e5596d7cf",name="myASG",reason="interrupt"} 1.0

Causes for your AutoScalingGroup instances scaling up/down can be viewed in the AWS EC2 Console -> AutoScalingGroups -> <AutoScalingGroupName> -> Activity

An example with common options:

# autoscaling groups to discover by tag
tags:
  - key: "kubernetes.io/cluster/my-cluster"
    value: "owned"
  - key: "eks:nodegroup-name"
    value: "my-spot-nodegroup"
# autoscaling group activity causes
causes:
  - pattern: ".*Spot Instance interruption notice.*"
    reason: "interrupt"
  - pattern: ".*EC2 instance rebalance recommendation.*"
    reason: "rebalance"
  - pattern: ".*user request.*"
    reason: "user_request"