Skip to main content

Artifact Registry and Continuous Integrations

Last updated on

Learn how to use Artifact Registry with the Continuous Integration (CI) module.

Build and push to Docker with Artifact Registry​

Harness CI offers a Build and push to Docker step that allows you to build and push a docker image to any registry. Usually, this requires a connector to the registry you want, but with AR you can connect to a registry directly, within the platform, without any complicated setup.

Upload Artifacts to Harness Artifact Registry​

Harness CI provides a native Upload Artifacts to Harness Artifact Registry step that simplifies the process of publishing artifacts directly to your Harness Artifact Registry. This dedicated step supports multiple package types and eliminates the need for complex scripting or external connectors.

CI Upload Artifact

The Upload Artifacts step allows you to configure artifact uploads through a simple interface. You can specify the package type, target registry, and source path for your artifacts. The source path typically points to artifacts generated from a previous step in your pipeline, such as a Run step that builds, creates, or downloads packages.

Source Path Configuration

Enter the Source Path where your artifact is located:

  • This path typically references the output from a previous step, such as:
    • /harness/target/*.jar for Maven artifacts
    • /harness/example.tar.gz for npm tarballs
    • /harness/build/ for Python wheels
  • You can use expressions to reference outputs from earlier steps

Package-specific requirements:

  • Generic packages: Requires explicit package name and version
  • Go packages: Requires explicit version specification
note

The Upload Artifacts step uses Harness CLI internally to upload artifacts to Harness Artifact Registry, ensuring consistent behavior and authentication across all package types.

Example YAML
pipeline:
  name: Upload Conda Package to Artifact Registry
  identifier: conda_upload_pipeline
  projectIdentifier: default_project
  orgIdentifier: default
  stages:
    - stage:
        name: Build and Upload
        identifier: build_upload_stage
        type: CI
        spec:
          cloneCodebase: true
          execution:
            steps:
              - step:
                  type: Run
                  name: Download Conda Package
                  identifier: download_conda_package
                  spec:
                    shell: Sh
                    command: |-
                      #!/bin/sh
                      set -e

                      # Configuration
                      PACKAGE_NAME="docutils-stubs"
                      CHANNEL="conda-forge"

                      echo "Fetching metadata for ${PACKAGE_NAME} from ${CHANNEL}"

                      # Fetch package metadata
                      JSON_DATA=$(curl -s "https://api.anaconda.org/package/${CHANNEL}/${PACKAGE_NAME}/files")

                      # Extract the first download URL
                      LATEST_URL=$(echo "${JSON_DATA}" \
                        | grep -o '"download_url": "[^"]*"' \
                        | head -n 1 \
                        | cut -d'"' -f4)

                      # Validate URL
                      if [ -z "${LATEST_URL}" ]; then
                        echo "Failed to retrieve download URL"
                        exit 1
                      fi

                      # Normalise protocol if required
                      case "${LATEST_URL}" in
                        //* ) LATEST_URL="https:${LATEST_URL}" ;;
                      esac

                      FILE_NAME=$(basename "${LATEST_URL}")

                      echo "Downloading ${FILE_NAME}"

                      # Download artifact
                      curl -L "${LATEST_URL}" -o "${FILE_NAME}"

                      # Resolve absolute path
                      FILE_PATH="$(pwd)/${FILE_NAME}"

                      echo "Resolved artifact path: ${FILE_PATH}"

                      # Write output using Harness supported contract
                      echo "FILE_PATH=${FILE_PATH}" >> "${DRONE_OUTPUT}"
              
              - step:
                  type: HarUpload
                  name: Upload to Artifact Registry
                  identifier: upload_to_registry
                  spec:
                    registryRef: my_conda_registry
                    packageType: CONDA
                    sourcePath: <+steps.download_conda_package.output.outputVariables.FILE_PATH>
          platform:
            os: Linux
            arch: Arm64
          runtime:
            type: Cloud
            spec: {}

Key Points:

  1. Step 1 (Run): Downloads a Conda package from conda-forge and outputs the file path using DRONE_OUTPUT
  2. Step 2 (HarUpload): References the output variable from Step 1 using <+steps.download_conda_package.output.outputVariables.FILE_PATH> and uploads to the specified registry
  3. Package Type: Set to CONDA to ensure proper indexing in Artifact Registry
  4. Source Path: Dynamically references the downloaded artifact location from the previous step

Add a Run or Plugin Step to Your CI Pipeline​

Seamlessly reference artifacts stored in Harness Artifact Registry within your CI Run and Plugin steps. This functionality eliminates the need for an external connector, providing a streamlined and user-friendly experience, while enabling efficient and simplified workflows.