Google Compute Engine (GCE)
- Add cloud
- Prepare your GCE account
- Add credentials
- Create a controller
- Next steps
Juju already has knowledge of the GCE cloud, which means adding your GCE account to Juju is quick and easy.
You can see more specific information on Juju’s GCE support (e.g. the supported regions) by running:
juju show-cloud google
To ensure that Juju’s information is up to date (e.g. new region support), you can update Juju’s public cloud data by running:
Although Juju knows how GCE works, there are a few tasks you must perform in order to integrate your account with Juju. We give an overview of the steps here:
- Using the CLI tools
- Assigning user permissions
- Managing service accounts
- Gathering credential information
- Enabling the Compute Engine API
The Google Cloud Platform Console (web UI) can also be used to complete the above steps.
The instructions on this page make use of the Identity and Access Management (IAM) framework to control access to your GCP account. Read Google’s Cloud IAM page for an overview.
We show how to use the Cloud SDK tools from Google to manage your GCP (Google Cloud Platform) account. The tools installation instructions presented here are for Ubuntu/Debian. See the link above for how to install on other platforms.
Install the tools in this way:
echo "deb http://packages.cloud.google.com/apt cloud-sdk main" \ | sudo tee -a /etc/apt/sources.list.d/google-cloud-sdk.list curl https://packages.cloud.google.com/apt/doc/apt-key.gpg | sudo apt-key add - sudo apt update && sudo apt install google-cloud-sdk
Now initialise the tool:
Among other things, you will be asked to enter a verification code in order to log in to GCP. This code is acquired by following a supplied hyperlink, which, in turn, will have you agree on the resulting page to allow Google Cloud SDK to access your Google account.
You will be given the choice of selecting an existing GCE project or of creating a new one. If creating, pick a unique name to prevent the command from exiting. If it does, re-invoke
gcloud init and choose option  to re-initialise.
When you’re done, try out the following commands (we created a project called ‘juju-gce-1’ during the init phase):
gcloud components list gcloud services list gcloud config list gcloud projects list gcloud projects describe juju-gce-1
See the gcloud command reference for more help with this tool.
Using the IAM framework, we’ll be associating credentials with our project at the Compute Engine service account level and not at the level of your Google user.
To download such credentials, however, your user (now known to the CLI tool) must have the authorisation to do so. This is done by assigning the role of ‘Service Account Key Admin’ to your user (insert your project ID and your user’s email address):
gcloud projects add-iam-policy-binding juju-gce-1 \ --member user:email@example.com \ --role roles/iam.serviceAccountKeyAdmin
A project’s service accounts are listed in this way:
gcloud iam service-accounts list --project juju-gce-1
You can create a new service account if:
- you are having trouble identifying an existing one to use
- your project does not yet have any service accounts
- you want one dedicated to Juju
Here, we will create a new one called ‘juju-gce-1-sa’:
gcloud iam service-accounts create juju-gce-1-sa \ --display-name "Compute Engine Juju service account"
For our example project, the list of service accounts is now:
NAME EMAIL Compute Engine Juju service account firstname.lastname@example.org
We must now give our chosen service account enough permissions so it can do what Juju asks of it. The roles of ‘Compute Instance Admin (v1)’ and ‘Compute Security Admin’ are sufficient:
gcloud projects add-iam-policy-binding juju-gce-1 \ --member serviceAccount:email@example.com \ --role roles/compute.instanceAdmin.v1 gcloud projects add-iam-policy-binding juju-gce-1 \ --member serviceAccount:firstname.lastname@example.org \ --role roles/compute.securityAdmin
Permissions can be configured in multiple ways due to the many IAM roles available. See upstream document Compute Engine IAM Roles for details.
Verify the roles now assigned to both your user and your service account:
gcloud projects get-iam-policy juju-gce-1
You are now ready to download credentials for your chosen service account. Here we’ve called the download file
gcloud iam service-accounts keys create juju-gce-1-sa.json \ --email@example.com
Store this file on the Juju client (e.g.
The section Using environment variables below explains where this data can be stored if you wish to use the
autoload-credentials command to add credentials to Juju.
The Compute Engine API needs to be enabled for your project but this requires your billing account to be first linked to your project.
Your billing (credit card) should have been set up when you registered with GCE. To see your billing account number:
gcloud alpha billing accounts list
ACCOUNT_ID NAME OPEN MASTER_ACCOUNT_ID 01ACD0-B3D759-187641 My Billing Account True
Use the account number/ID to link your project:
gcloud alpha billing projects link juju-gce-1 --billing-account 01ACD0-B3D759-187641
You can now enable the Compute Engine API for your project (this can take a few minutes to complete):
gcloud services enable compute.googleapis.com --project juju-gce-1
For some integrated Juju services (e.g. the Charmed Distribution of Kubernetes) it is useful to also enable the IAM management API:
gcloud services enable iam.googleapis.com --project juju-gce-1
Verify by listing all currently enabled services/APIs:
gcloud services list --project juju-gce-1
The Credentials page offers a full treatment of credential management.
In order to access Google GCE, you will need to add credentials to Juju. This can be done in one of three ways (as shown below).
The project that gets used by Juju is determined by the service account’s credentials used to create a controller. It is therefore recommended that the user-defined credential name strongly reflects the project name. This is chiefly relevant in a multi-project (multi-credential) scenario.
Alternately, you can use your credentials with Juju as a Service, where charms can be deployed within a graphical environment that comes equipped with a ready-made controller.
Armed with the gathered information, credentials can be added interactively:
juju add-credential google
The command will prompt you for information that the chosen cloud needs. An example session follows:
Enter credential name: juju-gce-1-sa Auth Types jsonfile oauth2 Select auth type [jsonfile]: Enter file: ~/.local/share/juju/juju-gce-1-sa.json Credential "juju-gce-1-sa" added locally for cloud "google".`
A YAML-formatted file, say
mycreds.yaml, can be used to store credential information for any cloud. This information is then added to Juju by pointing the
add-credential command to the file:
juju add-credential google -f mycreds.yaml
See section Adding credentials from a file for guidance on what such a file looks like.
With GCE you have the option of adding credentials using the following environment variable that may already be present (and set) on your client system:
In addition, a special variable may contain the path to a JSON-formatted file which, in turn, contains credential information:
Finally, on Linux systems, the file
$HOME/.config/gcloud/application_default_credentials.json may be used to contain credential data and is parsed by the above command as part of the scanning process. On Windows systems, the file is
Add this credential information to Juju in this way:
For any found credentials you will be asked which ones to use and what name to store them under.
For background information on this method read section Adding credentials from environment variables.
You are now ready to create a Juju controller for cloud ‘google’:
juju bootstrap google google-controller
Above, the name given to the new controller is ‘google-controller’. GCE will provision an instance to run the controller on.
For a detailed explanation and examples of the
bootstrap command see the Creating a controller page.
A controller is created with two models - the ‘controller’ model, which should be reserved for Juju’s internal operations, and a model named ‘default’, which can be used for deploying user workloads.
See these pages for ideas on what to do next:
Last updated 2 months ago.