Choosing which machines are managed

CostCapper looks for EC2 instances, RDS databases and Aurora clusters in your account's regions every 15 minutes. The Machines page lists them all.

Without a tag

You choose machines yourself: Manage or Stop managing on the Machines page. Machines launched later wait for you, so CostCapper never stops a new machine by surprise.

With a tag

Set a tag such as environment=dev in setup or on Settings. Every machine carrying it is managed and the rest are released (their open requests end). From then on, adding the tag to a machine in AWS manages it within 15 minutes, and removing the tag releases it. If you release a tagged machine yourself, CostCapper leaves it released.

If the appliance was installed with the CloudFormation option Scope = tagged, AWS only lets it touch machines tagged scheduler-managed=true, so that tag is fixed.

Machines that can't be managed

Some machines can't be switched off on request, and the Machines page says why: Auto Scaling group members (the group would replace them), spot instances, instances whose root disk is instance store (it loses its data when stopped), RDS read replicas and databases that have them, Aurora Serverless v1, and CostCapper itself.

The costcapper:managed tag

CostCapper tags each machine it manages costcapper:managed=true, and removes the tag when it stops managing one. The overnight failsafe uses it, and it shows you in the AWS console what CostCapper manages. It is the only tag CostCapper changes.

Request limits

Open a managed machine to set how long a request lasts by default (two hours) and the longest allowed (eight hours, at most seven days).