Step 2: Deploying Automation Suite to GCP

Important: Make sure to go through Step 1: Preparing the GCP deployment before proceeding.

Configuring the deployment

Once you prepare the GCP environment, you can directly deploy the infrastructure.

Take the following steps:

1. Download the templates from here and unzip them.
2. (Optional) Move the SSH keys generated at Step 1 in the templates folder.
3. The unzipped templates include a terraform.tfvars file with the parameters. Edit them according to your deployment.

Deployment parameters

Name	Description	Parameter type	Example
project_id	Project in which the resources will be deployed.	string	my-gcp-project
region	Region where the resources will be deployed.	string	europe-west1
zone	Default zone where zonal resources are deployed. The zone should belong to the region you chose.	string	europe-west1-c
installer_version	Automation Suite installer version.	string	2021.10.3
installer_url	Automation Suite installer URL. It overwrites the installer_version parameter and installs the version from the package.	string	https://download.uipath.com/automation-suite/installer.zip
perform_installation	Automatically install Automation Suite on the provided infrastructure.	bool	true
lb_fqdn	Fully qualified domain name (FQDN) for Automation Suite.	string	my-fqdn.net
ssh_user	Username used to log in VMs via SSH.	string	user
ssh_pub_key_path	Path to the SSH public key used to access VMs.	string	gcp_key.pub
ssh_private_key_path	Path to the SSH private key used to access VMs.	string	gcp_key
server_instance_type	Instance type of server machines.	string	n2-standard-32
agent_instance_type	Instance type of agents machines.	string	n2-standard-16
gpu_instance_type	(Optional) Parameter used only if the number of GPU agents is at least 1. Instance type of GPU agent machines.	string	n1-standard-16
gpu_type	(Optional) Parameter used only if the number of GPU agents is at least 1. GPU type for the GPU nodes.	string	nvidia-tesla-t4
server_count	Number of server VMs, must be odd	number	3
agent_count	Number of agent VMs.	number	1
gpu_agent_count	Number of GPU agent VMs	number	1
vm_custom_image	(Optional) Name of the custom image you want to use for the VM's. Leave empty to default to latest RHEL 8 GCP image.	string	my-rhel8-image
install_orchestrator	Install Orchestrator.	bool	true
install_aicenter	Install AI Center.	bool	true
install_documentunderstanding	Install Document Understanding.	bool	true
install_insights	Install Insights.	bool	true
install_data_service	Install Data Service	bool	true
install_test_manager	Install Test Manager.	bool	true
install_automation_ops	Install Automation Ops.	bool	true
install_automation_hub	Install Automation Hub.	bool	true
install_apps	Install Apps.	bool	true
install_action_center	Install Action Center.	bool	true
install_task_mining	Install Task Mining.	bool	true
install_as_robots	Install Automation Suite Robots.	bool	true
asrobots_agent_count	Instance type of ASobots dedicated agents.	number	1
asrobots_agent_instance_type	Number of dedicate ASRobots VMs.	string	n2-standard-16
install_process_mining	Install Process Mining.	bool	true
accept_license_agreement	By setting the parameter as true, you agree to the applicable license agreement.	bool	true

Advanced configuration

If you want to add any new configuration that is exposed via cluster_config.json, follow these steps:

1. Use your favorite text editor to edit the ./scripts/inputs_schema file from the unzipped package.
2. Change or add the json file as required.
   Note: Keep in mind that Terraform will replace ${var} variables defined in the metadata.tf with their respective values when altering the inputs_schema file.

Completing the deployment

Once you set the parameters, take the following steps:

1. Open a terminal and authenticate to GCP using the following command.

   gcloud auth application-default login

2. In the terminal, go to the folder where you unzipped the templates and run the following commands:

   terraform init
   terraform apply

3. Type Yes when asked for approval. The deployment will then run. It usually takes between 1 to 1.5 hours to complete.