Lokun

Lokun (Icelandic for 'shutdown') is a simple Google Cloud function to manage compute instances using a configured schedule.

Getting started

Configuring your compute instance

1. Go to your compute instance in Google Cloud Console and edit it
2. Add a network tag auto-shutdown
3. Configure the start/end time by adding a start-time and stop-time key/value to the custom metadata section. Times are in the HH:mm format and UTC-based.
4. If you want to configure different machine types on a daily basis, add a machineType-{day} key/value to the custom metadata section. When starting a machine, it will resize it if a different machine type is configured for that day. When no specific machine type is configured for that day, it will keep the same machine type.
   1. Use the 3-letter word for days: mon, tue, wed, thu, fri, sat and sun
   2. Use the machine types specified here: https://cloud.google.com/compute/docs/machine-types

An example configuration may look like:

This configuration results in the following schedule:

• Monday: a g1-small machine
• Tuesday till thursday: an n1-standard-1 machine
• Friday: a g1-small machine
• Saturday and sunday: an f1-micro machine

Creating a function

1. Go to the Google Cloud Console
2. Create a new Cloud Function
3. Give it a name, such as checkGcpInstances
4. Choose 128MB memory
5. Choose your function trigger (HTTP for example)
6. Choose 'Inline editor'
7. Grab index.js and package.json and copy-paste these into the corresponding editors
8. Use checkGcpInstances as function name to execute
9. Deploy your function and execute a test request

Triggering your function

Depending on your cloud function configuration you need to trigger the function at some interval. Google Cloud Functions provides different trigger mechanisms such as HTTP triggers, Cloud Pub/Sub and Cloud Storage buckets. When using an HTTP trigger you could configure an HTTP uptime/ping tool to call the function with an interval.

By default the Google Cloud project ID will be detected from the platform. You can also specify the project ID in the request body using the following payload:

{ "project": "PROJECT_ID" }

Authentication settings will also be detected from the platform.