Are you looking for some system that alerts when your today's spending on AWS exceeds some limit? That's just what this exporter is made for.
CloudWatch doesn't allow you to track costs in the same day, just past days or monthly predictions.
The exporter is a Python server that connects to AWS Cost Explorer with a customizable period, and exposes last responses as Prometheus metrics.
Configuration is made through environment variables:
| Environment variable | Description | Default |
|---|---|---|
| QUERY_PERIOD | Period to update metrics, querying AWS Cost Explorer API (0.01$ per request) | 1800 |
| METRIC_TODAY_DAILY_COSTS | Enable aws_today_daily_costs metric | Not set |
| METRIC_YESTERDAY_DAILY_COSTS | Enable aws_yesterday_daily_costs metric | Not set |
| METRIC_TODAY_DAILY_USAGE | Enable aws_today_daily_usage metric | Not set |
| METRIC_TODAY_DAILY_USAGE_NORM | Enable aws_today_daily_usage_norm metric | Not set |
| AWS_FILTER | User AWS Cost Explorer Filter | Not set |
| CONFIG_PATH | filter config filepath | /cost-exporter |
{
"Version": "2012-10-17",
"Statement": [
{
"Sid": "VisualEditor0",
"Effect": "Allow",
"Action": "ce:*",
"Resource": "*"
}
]
}
Run on localhost
docker run -e METRIC_TODAY_DAILY_COSTS=yes -p 5000:5000 nachomillangarcia/prometheus_aws_cost_exporter:latest
Run on public machine
docker run -e METRIC_TODAY_DAILY_COSTS=yes -p 5000:5000 nachomillangarcia/prometheus_aws_cost_exporter:latest --host 0.0.0.0
Use the provided Helm Chart in this repository to add this service to your Kubernetes cluster
cd helm
helm install -e METRIC_TODAY_DAILY_COSTS=yes --name prometheus-aws-cost-exporter --namespace kube-system .
To grant necessary AWS IAM permissions, I suggest to use kube2iam. Pod annotations are customizable for this chart, check the README.md in helm folder