scaler's Introduction

Scaler

Problem

Scale cluster at pre-defined schedule e.g at night and weekends to save cost.

Scaler

Scaler can modify auto scaling groups to change cluster size at pre-defined schedule

Usage

Scaler can be used to modify max, min and desired capacity of an auto scaling group.

Scaler supports the following flags

Short Flag	Long Flag	Type	Description
-d	--desired	int	Desired no of instances (Required)
-h	--help		Help for Scaler
-m	--max	int	Maximum no of instances (Required)
-i	--min	int	Minimum no of instances (Required)
-p	--provider	string	Cloud provider to user. Valid Values (aws)
-r	--region	string	Region in which auto scaling group exists (Required)
-a	--roleArn	string	Arn of role to assume (Required)
-n	--scalerName	string	Name of Auto Scaling group (Required)

Run

Scale can be run by passing arguments to the published image. e.g

docker run -it docker.io/stakater/scaler:0.0.1 --roleArn arn:aws:iam::449074299682:role/nodes.stackator.com --region us-west-2 --max 0 --min 0 --desired 0 --provider aws --scalerName nodes.stackator.com

Use Case

Scaler can be used in combination with Cronjob to control the cluster size during different times. e.g You can bring down your cloud cost by keeping the servers shut down on weekends

A sample configuration of kubernetes cronjob can be found here.

Adding Support for Cloud Providers

Provider can be implemented to modify auto scaling groups on other cloud providers e.g Azure, Google Cloud etc. If new parameters are needed then they can be added by modifying Scaler Options struct in common.go. The values will automatically be available in the Init method of new implementation of Provider