Use Reliably Environments

This guide goes into the details of creating and using Reliably environments to inject contexts into your Reliably Plans.

Reliably Experiments describe a set of activities to conduct. These activities are usually context-dependent. For instance, a given experiment may be designed to work against Kubernetes. You would expect that such an experiment can target any Kubernetes cluster that you run, therefore credentials should not be embedded into the experiment.

This is where Reliably Environments come into play. They allow you to declare a set of environment variables and secrets that you can attach to a Reliably Plan for a particular run of an Experiment. Thus providing the context it needs.

The Basics

Create an Environment

Reliably Environments are only used by Reliably Cloud Deployment type. For other types of deployments, the environment responsibility is held by the platform where you execute the Plan. For instance, a GitHub deployment will use GitHub Environments rather than Reliably Environments. On a CLI deployment, you will pass the context via environment variables your platform supports directly.

To create an environment, go to the New environment page and start filling out the form as needed.

Provide a name that you can easily recall what it covers later on. Then set as many environment variables and secrets, which are encrypted internally, as you need.

When creating a secret as a file, the path must start with /home/svc.

Delete an Environment

You can delete environments at will but only when all Reliably Plan that use them have also been deleted first. This prevents mistakes where a Plan tries to run and cannot find its Environment.

Common Patterns

This section will introduce common Environment patterns that will help you pass the right context to a variety of experiments.

These patterns apply to any Reliably Deployment type. But you declare these variables and secrets in Reliably only when using a Reliably Cloud deployment. Otherwise, these variables and secrets are declared directly on the platform you use to execute Reliably Plans.


Experiments targetting Kubernetes clusters usually require a service account that will define the Kubernetes API server endpoint and credentials to authenticate to it. The service account should then have the right roles to perform the experiment’s activities.

Environment VariableKUBECONFIG/home/svc/.kube/config
Secret File/home/svc/.kube/configThe content of a Kubernetes Service Account file


Experiments targeting AWS resources usually require enough information to authenticate.

Environment VariableAWS_REGIONA valid AWS region
Secret File/home/svc/.aws/credentialsThe content of an shared credentials file

Alternatively, you can only use environment variables as well:

Environment VariableAWS_REGIONA valid AWS region
Secret VariableAWS_ACCESS_KEY_IDThe AWS access key for the AWS account to use
Secret VariableAWS_SECRET_ACCESS_KEYThe AWS secret key for the AWS account to use

Google Cloud Platform

Experiments targetting GCP services usually require a service account, which should have the right roles to perform the experiment’s activities.

Environment VariableGOOGLE_APPLICATION_CREDENTIALS/home/svc/gcp.json
Secret File/home/svc/gcp.jsonThe content of a GCP Service Account key file


Experiments targetting GitHub services require a token with the appropriate permissions for the experiments in the Reliably Plan.

Secret VariableGITHUB_TOKENThe token value