Interactive Terraform Enterprise Installation

Delivery

This document outlines the procedure for using the Terraform Enterprise installer to set up Terraform Enterprise on a customer-controlled machine.

Proxy Usage

If your installation requires using a proxy server, you will be asked for the proxy server information when you first run the installer via ssh. This proxy server will be used for all outbound HTTP and HTTPS connections.

Optionally, if you're running the installer script in an automated manner, you can pass a http-proxy flag to set the address of the proxy. For example:

./install.sh http-proxy=http://internal.mycompany.com:8080

Proxy Exclusions (NO_PROXY)

If certain hostnames should not use the proxy and the instance should connect directly to them (for instance, for S3), then you can pass an additional option to provide a list of domains:

./install.sh additional-no-proxy=s3.amazonaws.com,internal-vcs.mycompany.com,example.com

Passing this option to the installation script is particularly useful if the hostnames that should not use the proxy include services that the instance needs to be able to reach during installation, such as S3. Alternately, if the only hosts you need to add are those that are not used during installation, such as a private VCS instance, you can provide these hosts after initial installation is complete, in the settings tab in your dashboard (available on port 8800 under /console/settings).

Reconfiguring the Proxy

To change the proxy settings after installation, use the Console settings page, accessed from the dashboard on port 8800 under /console/settings.

On the Console Settings page, there is a section for HTTP Proxy:

This change updates the proxy settings for the Terraform Enterprise application services. To update the proxy settings for the installer (for example, to handle configuration tests correctly), additional steps are necessary:

1. Locate the Replicated configuration files on the instance under either /etc/sysconfig/ or /etc/default: replicated and replicated-operator.
2. Open the files for editing. On the line that includes REPLICATED_OPTS for replicated or REPLICATED_OPERATOR_OPTS for replicated-operator, add -e HTTP_PROXY=<your proxy url> -e NO_PROXY=<list of no_proxy hosts> to the existing command options. The list of no_proxy hosts is a comma-separated list with no spaces, and should include following addresses 127.0.0.1,<DOCKER0 INTERFACE IP>,<IP ADDRESS OF TFE INSTANCE>,<HOSTNAME OF TFE INSTANCE> but not limited to.
3. Docker also needs to be able to communicate to endpoints with the same rules of proxy settings as replicated and replicated-operator, the steps 1-6 of this document are required. NOTE: Please take precautions on application outage when applying configuration change, i.e. wait for all runs to finish, prevent new runs to trigger
4. Restart the Replicated services following the instructions for your distribution.

Proxy and Service Discovery Process

Unless the execution mode of a workspace is set to "local", Terraform Enterprise performs remote operations, running Terraform in its own worker VMs.

When running within Terraform Enterprise's worker VMs, Terraform uses service discovery to find the Terraform Enterprise service itself. Depending on your infrastructure setup, you may need to tell Terraform Enterprise not to access its own hostname via the proxy, so that Terraform can communicate with the Terraform Enterprise services.

To do this, add Terraform Enterprise's fully qualified hostname to the Proxy Bypass setting in Terraform Enterprise's dashboard. The proxy bypass setting can be found on port 8800 on the path /settings.

Save the configuration using the Save button at the bottom of the page. Once configuration has been saved, please proceed to restart the application.

You can use the equivalent setting, extra_no_proxy, to automate this step.

The Terraform CLI performs a TLS handshake with Terraform Enterprise during service discovery, and will need access to the Certificate Authority that Terraform Enterprise uses.

Either add the CA to the CA Custom Bundle configuration so that Terraform Enterprise will make the CA available in the worker container, or, if a custom worker image is configured, add the CA directly to its certificate file (located at /etc/ssl/certs/ca-certificates.crt).

TLS Configuration

There are two sections for TLS configuration; the "TLS Key & Cert" section and the "SSL/TLS Configuration" section.

TLS Key & Cert

The "TLS Key & Cert" section is where the TLS certificate and private key can be configured to allow HTTPS connections to Terraform Enterprise. The TLS certificate and private key files can be self-signed, located in a path on the server, or uploaded. Both the TLS certificate and private key files must be PEM-encoded. The TLS certificate file can contain a full chain of TLS certificates if necessary.

For convenience, a brand new Terraform Enterprise installation may prompt for these settings after the initial setup. You can provide a key and certificate immediately, or use a self-signed certificate to begin with and change the settings later.

For an existing installation, these settings can be found in the Replicated console on port 8800. Click on the gear icon in the top right corner, click "Console Settings", and scroll to the "TLS Key & Cert" section.

The key and certificate settings can be one of three values, each of which are detailed below.

Self-signed (generated)

When the "Self-signed (generated)" radio button is selected, a self-signed TLS certificate and private key will be automatically generated. An example screenshot is below:

Server path

When the "Server path" radio button is selected, the TLS certificate and private key will be read from the specified file paths on the server. An example screenshot is below:

Upload files

When the "Upload file" radio button is selected, the TLS certificate and private key must be uploaded. An example screenshot is below:

Certificate Authority (CA) Bundle

Terraform Enterprise needs to be able to access all services that it integrates with, such as VCS providers or database servers. Because it typically accesses them via SSL/TLS, it is critical that the certificates used by any service that Terraform Enterprise integrates with are trusted by Terraform Enterprise.

This section is used to allow Terraform Enterprise to connect to services that use SSL/TLS certificates issued by private CAs. It allows multiple certificates to be specified as trusted, and should contain all certificates that Terraform Enterprise should trust when presented with them from itself or another application.

A collection of certificates for trusted issuers is known as a Certificate Authority (CA) Bundle. All certificates in the certificate signing chain, meaning the root certificate and any intermediate certificates, must be included here. These multiple certificates are listed one after another in text format.

Certificates must be formatted using PEM encoding, that is, as text. For example:

-----BEGIN CERTIFICATE-----MIIFtTCCA52gAwIBAgIIYY3HhjsBggUwDQYJKoZIhvcNAQEFBQAwRDEWMBQGA1UEAwwNQUNFRElDT00gUm9vdDEMMAoGA1UECwwDUEtJMQ8wDQYDVQQKDAZFRElDT00xCzAJBgNVBAYTAkVTMB4XDTA4MDQxODE2MjQyMloXDTI4MDQxMzE2MjQyMlowRDEWMBQGA1UEAwwNQUNFRElDT00gUm9vdDEMMAoGA1UECwwDUEtJMQ8wDQYDVQQKDAZF....-----END CERTIFICATE----------BEGIN CERTIFICATE-----MIIB5zCCAY6gAwIBAgIUNJADaMM+URJrPMdoIeeAs9/CEt4wCgYIKoZIzj0EAwIwUjELMAkGA1UEBhMCVVMxCzAJBgNVBAgTAkNBMRYwFAYDVQQHEw1TYW4gRnJhbmNpc2NvMR4wHAYDVQQDExVoYXNoaWNvcnAuZW5naW5lZXJpbmcwHhcNMTgwMjI4MDYx....-----END CERTIFICATE-----

The user interface to upload these certificates looks like this:

TLS Versions

As of version 201902-01, TLS versions 1.0 and 1.1 are no longer supported in Terraform Enterprise. Your options now include TLS v1.2 and TLS v1.3:

TLS Ciphersuites

Terraform Enterprise uses the HIGH:!aNULL:!MD5 ciphersuites by default. You can also provide a custom ciphersuite if necessary.

(This value must be defined in the OpenSSL cipher list format

Force TLS with HSTS

Terraform Enterpise may be configured to force TLS via the enabling of HTTP Strict Transport Security:

Custom Image

Terraform Enterprise performs Terraform runs in ephemeral containers, using a built-in container image by default. To add custom tools or logic to your Terraform run environment, you must build a custom image and configure Terraform Enterprise to use it.

The process for building a custom image differs depending on whether your run_pipeline_mode setting is set to agent or legacy. As of v202302-1, the default run_pipeline_mode is agent.

See the release notes and the worker to agent migration guide for more information about the run pipeline mode change.

Agent

Build your custom image using the Dockerfile below. Then update the custom_agent_image_tag setting with your image (e.g. registry.example.com/example/tfc-agent:custom-tag).

Requirements

• The base image must be hashicorp/tfc-agent:1.6.0 or later.

Dockerfile

FROM hashicorp/tfc-agent:latest # Switch the to root user in order to perform privileged actions such as# installing software.USER root # Install sudo. The container runs as a non-root user, but people may rely on# the ability to apt-get install things.RUN apt-get -y install sudo # Permit tfc-agent to use sudo apt-get commands.RUN echo 'tfc-agent ALL=NOPASSWD: /usr/bin/apt-get , /usr/bin/apt' >> /etc/sudoers.d/50-tfc-agent # Switch back to the tfc-agent user as needed by Terraform agents.USER tfc-agent

Legacy

Build your custom image using either the Ubuntu or Red Hat Dockerfile below. Then update the custom_image_tag setting with your image (e.g. registry.example.com/example/terraform-enterprise-worker:custom-tag).

Requirements

• The base image must be ubuntu:bionic or an offical Red Hat Enterprise Linux 7 image (e.g. registry.access.redhat.com/ubi7/ubi-minimal:latest).
• The image must exist on the Terraform Enterprise host. You can add it by running docker pull from a local registry or any other similar method.
• The software packages defined in the examples below must be installed on the image.
• All necessary CA certificates must be added to the container's CA bundle. The CA certificates configured in the CA Bundle settings will not be automatically added to this image at runtime. See the examples below for more details.
• Terraform must not be installed on the image. Terraform Enterprise installs Terraform at runtime.

Ubuntu

FROM ubuntu:bionic # Install required software for Terraform Enterprise.RUN DEBIAN_FRONTEND=noninteractive && \    apt-get update && \    apt-get install -y --no-install-recommends awscli ca-certificates curl daemontools git-core iproute2 netcat-openbsd openssh-client psmisc redis-tools ssh sudo unzip wget # Include all necessary CA certificates.ADD example-root-ca.crt /usr/local/share/ca-certificates/ADD example-intermediate-ca.crt /usr/local/share/ca-certificates/RUN update-ca-certificates

Red Hat Enterprise Linux 7

In addition to the packages that yum installs, curl installs the envdir tool Python port because it is a dependency of the Terraform Build Worker service. If the envdir tool is not installed on the alternative worker image, Terraform runs will fail within Terraform Enterprise.

FROM registry.access.redhat.com/ubi7/ubi-minimal:latest # Update installed packages and clear cache.RUN microdnf --assumeyes update && \    rm --recursive --force /var/cache/yum # Install required software for Terraform Enterprise.RUN microdnf --assumeyes install curl git iproute nmap-ncat openssh openssl psmisc sudo unzip wget && \    microdnf clean all && \    curl --location --output /usr/local/bin/envdir https://github.com/jezdez/envdir/releases/download/0.7/envdir-0.7.pyz && \    chmod +x /usr/local/bin/envdir # Include all necessary CA certificates.ADD example-root-ca.crt /usr/share/pki/ca-trust-source/anchorsADD example-intermediate-ca.crt /usr/share/pki/ca-trust-source/anchorsRUN update-ca-trust

Executing Custom Scripts

The custom image supports executing custom scripts during different points of a run.

Initialize Script

To execute an initialize script, ensure your worker image contains an executable shell script at /usr/local/bin/init_custom_worker.sh. This script, and all commands it invokes, will be executed before a Terraform Enterprise run executes terraform init. This initialize script will be executed during both plans and applies.

Example Dockerfile snippet for adding an initialize script:

ADD init_custom_worker.sh /usr/local/bin/init_custom_worker.sh

Finalize Script

To execute a finalize script, ensure your worker image contains an executable shell script at /usr/local/bin/finalize_custom_worker.sh. This script, and all commands it invokes, will be executed after a Terraform Enterprise run finishes executing terraform plan or terraform apply. This finalize script will be executed during both plans and applies.

Example Dockerfile snippet for adding a finalize script:

ADD finalize_custom_worker.sh /usr/local/bin/finalize_custom_worker.sh

Operational Mode Decision

Terraform Enterprise can store its state in a few different ways, and you'll need to decide which works best for your installation. Each option has a different approach to recovering from failures. The mode should be selected based on your organization's needs. See Pre-Install Checklist: Operational Mode Decision for more details.

Installation

The installer can run in two modes, Online or Airgapped. Each of these modes has a different way of executing the installer, but the result is the same.

Run the Installer - Online

If your instance can access the Internet, use the Online install mode.

1. From a shell on your instance:
   • To execute the installer directly, run curl https://install.terraform.io/ptfe/stable | sudo bash.
   • To inspect the script locally before running, run curl https://install.terraform.io/ptfe/stable > install.sh and, once you are satisfied with the script's content, execute it with sudo bash install.sh.
     • RedHat Enterprise Linux requires Docker to be pre-installed. As such, execute the installer script using sudo bash install.sh no-docker to prevent the installer script from automatically installing Docker.
2. The installation will take a few minutes and you'll be presented with a message about how and where to access the rest of the setup via the web. This will be https://<TFE HOSTNAME>:8800.
   • You will see a security warning when first connecting. This is expected and you'll need to proceed with the connection anyway.

Run the Installer - Airgapped

If the instance cannot reach the Internet, follow these steps to begin an Airgapped installation.

Prepare the Instance

1. Airgap installations require Docker to be pre-installed. Double-check that your instance has a supported version of Docker (see Pre-Install Checklist: Software Requirements for details).
2. Download the .airgap file using the information given to you in your setup email and place that file somewhere on the the instance.
   • If you use are using wget to download the file, be sure to use wget --content-disposition "<url>" so the downloaded file gets the correct extension.
   • The url generated for the .airgap file is only valid for a short time, so you may wish to download the file and upload it to your own artifacts repository.
3. Download the installer bootstrapper and put it into its own directory on the instance (e.g. /opt/tfe-installer/)
   • The downloaded file may be named replicated.tar.gz or latest.tar.gz. We'll refer to it as replicated.tar.gz for the rest of this page.

Execute the Installer

From a shell on your instance, in the directory where you placed the replicated.tar.gz installer bootstrapper:

1. Run tar xzf replicated.tar.gz
2. Run sudo ./install.sh airgap
3. When asked, select the interface of the primary private network interface used to access the instance.
4. The software will take a few minutes and you'll be presented with a message about how and where to access the rest of the setup via the web. This will be https://<TFE HOSTNAME>:8800.
   • You will see a security warning when first connecting. This is expected and you'll need to proceed with the connection anyway.

Continue Installation In Browser

1. Configure the hostname and the SSL certificate.
2. Upload the license file provided to you in your setup email.
3. Indicate whether you're doing an Online or Airgapped installation. Choose Online if you're not sure.
   • If you are doing an Airgapped install, provide the path on the the instance to the .airgap file that you downloaded using the initial instructions in your setup email.
4. Secure access to the installer dashboard. We recommend at least setting up the simple password authentication. If you're so inclined, LDAP authentication can also be configured.
5. The system will now perform a set of pre-flight checks on the instance and the configuration up to this point and indicate any failures. You can either fix the issues and re-run the checks, or ignore the warnings and proceed. If the system is running behind a proxy and is unable to connect to releases.hashicorp.com:443, it is likely safe to proceed; this check does not currently use the proxy. For any other issues, if you proceed despite the warnings, you are assuming the support responsibility.
6. Set an encryption password used to encrypt the sensitive information at rest. The default value is auto-generated, but we strongly suggest you create your own password. Be sure to retain the value, because you will need to use this password to restore access to the data in the event of a reinstall. See Encryption Password for more information.
7. Configure the operational mode for this installation. See Pre-Install Checklist: Operational Modes for information on what the different values are. Ensure that you've met the relevant pre-install requirements for the mode you chose.
8. Optional: Adjust the concurrent capacity of the instance. This should only be used if the hardware provides more resources than the baseline configuration and you wish to increase the work that the instance does concurrently. This setting should be adjusted with care as setting it too high can result in an very unresponsive instance.
9. Optional: Provide the text version of a certificate (or certificates) that will be added to the trusted list for the product. This is used when services the product communicates with do not use globally trusted certificates but rather a private Certificate Authority (CA). This is typically used when Terraform Enterprise uses a private certificate (it must trust itself) or a VCS provider uses a private CA.
10. Optional: Adjust the path used to store the vault files that are used to encrypt sensitive data. This is a path on the host system, which allows you to store these files outside of the product to enhance security. Additionally, you can configure the system not to store the vault files within any snapshots, giving you full custody of these files. These files will need to be provided before any snapshot restore process is performed, and should be placed into the path configured.
11. Optional: Configure the product to use an externally managed Vault cluster. See Externally Managed Vault Cluster for details about the required Vault configuration before using this option.

Finish Bootstrapping

Once configured, the software will finish downloading. When it’s ready, the UI will indicate so and there will be an Open link to click to access the Terraform Enterprise UI.

Configuration

After completing a new install you should head to the configuration page to continue setting up Terraform Enterprise.