Atlantis PR Automation
You can integrate Terrascan into an Atlantis setup using one of the two ways:
- Method 1: Atlantis Workflow-based integration which sends scan requests to an independently running terraform server
- Method 2: Custom Atlantis container image which has an integrated Terrascan
In either scenario, the configuration of Atlantis varies from installation to installation. For instructions to install, configure, and use Atlantis, see the Atlantis documentation.
Method 1: Workflow-based integration
In this method, you can modify or create a custom workflow for Atlantis so your repositories will be scanned by Terrascan as part of the pull request automation.
The following requirements must be met before starting the integration workflow:
- The atlantis server must have TCP connectivity to where the terrascan server is running.
curlcommand must be installed on the system so the
terrascan-remote-scan.shscript can make the scan request. Atlantis’s docker image has curl preinstalled.
Integration steps for Workflow based integration
- Modify Workflow
- Configure the Script
- Run Atlantis
- Modify your workflow to call
terrascan-remote-scan.shduring the plan stage.
- See the ‘plan’ detailed below:
- the first three
run: terraformcommands are the default for an atlantis workflow.
Note: The values for the variables
$PLANFILEreferenced in the second and third run commands in the yaml below are automatically provided by atlantis
- The fourth
run terrascan-remote-scan.shinitiates the Terrascan scan request.
Note: By default, the
terrascan-remote-scan.shscript can be found under the
scriptsdirectory in this project; copy this to a location where it can be executed by the Atlantis server. If the
terrascan-remote-scan.shscript is not in the directory where the Atlantis server command is being run to, you will have to specify the path to the script in the fourth run command.
repos: - id: /.*/ workflow: terrascan workflows: terrascan: plan: steps: - run: terraform init -input=false -no-color - run: terraform workspace select -no-color $WORKSPACE - run: terraform plan -input=false -refresh -no-color --out $PLANFILE - run: terrascan-remote-scan.sh
terrascan-remote-scan.sh script according your environment. The script is located here. Open the script with your any editor of your choice and review the following six settings which is found at the top of the file:
TERRASCAN_SERVER=192.168.1.55 TERRASCAN_PORT=9010 IGNORE_LOW_SEVERITY=false IAC=terraform IAC_VERSION=v14 CLOUD_PROVIDER=aws
Descriptions of these settings are as follows:
TERRASCAN_SERVERis the hostname or IP address of the host running the terrascan server. This will be used by the script to submit the scan request.
TERRASCAN_PORTis the TCP port which Terrascan server is listening on. By default, this is
IGNORE_LOW_SERVERITYallows you to specify the scan response for low-severity findings in the code. During a scan if the
terrascan-remote-scan.shshould fail a build if a low-severity finding is found. Some users will want to set this to
trueso they may ignore low-severity findings.
CLOUD_PROVIDERare terrascan options. Descriptions and valid values can be found by running
terrascan scan -h.
Run Atlantis with the
terrascan-workflow.yaml as a server-side repo configuration. The command for this depends on how you choose to deploy Atlantis.
If running the Atlantis binary directly, use the following command:
$ atlantis server \ --atlantis-url="$URL" \ --gh-user="$USERNAME" \ --gh-token="$TOKEN" \ --gh-webhook-secret="$SECRET" \ --repo-allowlist="$REPO_ALLOWLIST" \ --repo-config=terrascan-workflow.yaml
Note: The variables in the example above must be configured separately using
exportor similar shell methods.
Important: Before the first pull request is processed, run Terrascan in
server mode using the following command:
Automated scanning and results
When the systems are running, if Atlantis is initiated either via a pull request, or via a comment of
atlantis plan, Terrascan will be also be invoked as part of the atlantis plan flow. Scan results are reported as part of the pull request as comments, this notifies the reviewers before approving a requests. If any issues are found the test will be marked as failed.
Method 2: Custom Atlantis Container
Terrascan offers a custom container built on top of the official Atlantis container image, which allows users to run IaC scans with Terrascan, in addition to the usual atlantis usage. There’s a built-in atlantis workflow configured inside the
container which is ready to be used.
workflow.yaml file used is the
atlantis/workflow.yaml in the Terrascan repo.
Alternatively, you can also override that default workflow using the
Steps to use the custom Atlantis container
In code repository, usage is exactly the same as atlantis, add a comment
atlantis plan and
atlantis plan to your Pull Requests to trigger the custom atlantis-terrascan workflow.
To use the default built-in container image:
docker pull tenable/terrascan_atlantis
To build your own container image:
docker build ./integrations/atlantis -t <image_name>
Run the container:
docker run \ --env-file=<.env-file> \ -p 4141:4141 \ -v <pwd>/config_data/:/etc/terrascan/ \ tenable/terrascan_atlantis server \ --gh-user="$USERNAME" --gh-token="$TOKEN" --gh-webhook-secret="$SECRET" \ --repo-allowlist="$REPO_ALLOWLIST" \ -c /etc/terrascan/config.toml
The syntax of the Atlantis server command here is same as in atlantis docs, except for an optional
-c flag which can be used to specify the file path for the toml config to be used by Terrascan. Another way to provide the toml config filepath would be the TERRASCAN_CONFIG environment variable. You need to provide all the environment variables that terraform requires to operate with your respective cloud providers.
Note: As a good practice, Terrascan recommends use of a specific tag of the container image rather than the latest tag.
Running a scan
With everything configured, a local Terrascan scan will be triggered as part of the Atlantis plan workflow.
Was this page helpful?
Glad to hear it! Please tell us how we can improve.
Sorry to hear that. Please tell us how we can improve.