Spot Price Monitor

Table of contents

• Introduction
• Usage
• Related
• Communication
• Contributing
• License

Introduction

Reads instance types and availability zones from Kubernetes labels to expose as Prometheus metrics, the current spot prices of spot instances within a cluster.

Usage

Deploy to Kubernetes

A docker image is available at quay.io/pusher/k8s-spot-price-monitor. These images are currently built on pushes to master. Releases will be tagged.

Sample Kubernetes manifests are available in the deploy folder.

To deploy in clusters using RBAC, please apply all of the manifests (Deployment, ClusterRole, ClusterRoleBinding and ServiceAccount) in the deploy folder but uncomment the serviceAccountName in the Deployment.

Requirements

For the K8s Spot Price Monitor to filter spot instances as expected; you will need an identifying label on your spot instances.

We add a label node-role.kubernetes.io/spot-worker to our spot instances and hence this is the default for the spot-label flag.

To achieve this, add the following flag to your Kubelet:

--node-labels="node-role.kubernetes.io/spot-worker=true"

Since the script uses a built-in, well-known label for looking up instance types (beta.kubernetes.io/instance-type), this project supports K8s v1.7+.

IAM

To fetch Spot Prices, the Spot Price Monitor will need the following IAM role policy.

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Sid": "Stmt1519813867626",
      "Action": [
        "ec2:DescribeSpotPriceHistory"
      ],
      "Effect": "Allow",
      "Resource": "*"
    }
  ]
}

Product descriptions

For EC2 Classic accounts the default Linux/UNIX product description will only work for non-VPC instance types, and for VPC accounts it will only work for VPC instance types.

It's possible to override the product descriptions with the -p/--products flag to work around this:

spot-price-monitor.py --products "Linux/UNIX" "Linux/UNIX (Amazon VPC)"

Flags

usage: spot-price-monitor.py [-h] [--running-in-cluster RUNNING_IN_CLUSTER]
                             [-l SPOT_LABEL] [-i SCRAPE_INTERVAL]
                             [-m METRICS_PORT] [-r REGION]
                             [-p PRODUCTS [PRODUCTS ...]]

Monitors kubernetes for spot instances and exposes the current spot prices as
prometheus metrics

optional arguments:
  -h, --help            show this help message and exit
  --running-in-cluster RUNNING_IN_CLUSTER
                        Will load kubernetes config from the pod environment
                        if running within the cluster, else loads a kubeconfig
                        from the running environemnt (Default: False)
  -l LABEL, --spot-label LABEL
                        Specifies the label applied to all spot instances in
                        the cluster to identify these from other types of
                        instances (Default: node-role.kubernetes.io/spot-
                        worker)
  -i SCRAPE_INTERVAL, --scrape-interval SCRAPE_INTERVAL
                        How often (in seconds) should the prices be scraped
                        from AWS (Default: 60)
  -m METRICS_PORT, --metrics-port METRICS_PORT
                        Port to expose prometheus metrics on (Default: 8000)
  -r REGION, --region REGION
                        The region that the cluster is running in (Default:
                        us-east-1)
  -p PRODUCTS [PRODUCTS ...], --products PRODUCTS [PRODUCTS ...]
                        List of product (descriptions) to use for filtering
                        (Default: Linux/UNIX)

Related

• K8s Spot Rescheduler: Move nodes from on-demand instances to spot instances when space is available.
• K8s Spot Termination Handler: Gracefully drain spot instances when they are issued with a termination notice.

Communication

• Found a bug? Please open an issue.
• Have a feature request. Please open an issue.
• If you want to contribute, please submit a pull request

Contributing

Please see our Contributing guidelines.

License

This project is licensed under Apache 2.0 and a copy of the license is available here.