Sigrid on-premise: Analysis configuration

This documentation covers on-premise Sigrid. It is not applicable for cloud-based Sigrid.

Prerequisites

Your development platform will need access to the Sigrid on-premise Docker containers.

Each system to be analyzed needs an analysis configuration in the form of a file called sigrid. yaml in the root directory of the system. Typically, this configuration is maintained by the developers responsible for the system and consequently is not discussed here. Developers are referred to the analysis configuration reference.

Sigrid’s analyses require access to an S3-compatible object store. This can be Amazon’s implementation, or an on-premise equivalent that supports Amazon’s S3 API, such as MinIO or Ceph.

Sigrid’s analysis image includes s5cmd to access the object store, which is based on the official Amazon S3 SDK. Because it is based on the official Amazon S3 SDK, it supports all authentication methods that Amazon provides. Typically, in a CI/CD environment, the AWS S3 SDK uses environment variables to hold an access key. Consequently, typically the following environment variables need to be set:

• AWS_ACCESS_KEY_ID
• AWS_SECRET_ACCESS_KEY
• AWS_DEFAULT_REGION

Configuring pipeline jobs

For on-premise solutions, we expect that source code is analyzed in a job in a CI/CD pipeline. In this document, we use GitLab as the example CI/CD environment. As analysis is based on a Docker image, any CI/CD environment that can run Docker containers will do.

The following GitLab job illustrates how to run an analysis:

sigrid-publish:
  image:
    # Pulls from the private part of SIG's registry at Docker Hub; you may need to log in first, or replace this with the image name as cached in your internal registry:
    name: "softwareimprovementgroup/sigrid-multi-analyzer:1.0.20241206"
  variables:
    # These are all environment variables. For defaults, see the table below.
    # Note that typically, all environment variables marked as "shared" in the table
    # below would be set globally in the CI/CD environment:
    CUSTOMER: "company_name"
    SYSTEM: "$CI_PROJECT_NAME"
    SIGRID_URL: "https://sigrid.my-company.com"
    SIGRID_CI_TOKEN: "secret"
    S3_ENDPOINT: "https://minio.my-company.com"
    BUCKET: "some-bucket"
    AWS_ACCESS_KEY_ID: "some-id"
    AWS_SECRET_ACCESS_KEY: "also-secret"
    AWS_REGION: "us-east-1"
    SIGRID_SOURCES_REGISTRATION_ID: "gitlab-onprem"
  script:
    - "run-analyzers --publish"

Note that the image name contains an explicit Docker image tag (1.0.20241206 in this example). It is important that the tag matches the tags used in Sigrid’s Helm chart: all components of Sigrid must always use the same version. SIG recommends using an environment-wide variable instead of hardcoding the tag.

In GitLab, a CI/CD pipeline job with an image property starts the named Docker image, mounts a directory into it where it (automatically) checks out the source code of the current project, and runs the command(s) provided in the script tag. Other CI/CD environments provide a similar structure, although details may differ:

• Start a container.
• Ensure the source code of the project is available in it.
• Run the provided script inside the container (thus overriding the image entrypoint).

The run-analyzers script takes one optional command line parameter:

• --publish: run all analyses, persist analysis results in Sigrid, show analysis results on stdout and set exit code.
• --publishonly: run all analyses and persist analysis results in Sigrid.
• None: run analyses but do not persist analysis results (only show analysis results on stdout and set exit code).

Sigrid CI environment variables

Sigrid CI is configured with environment variables. The following table lists all environment variables with their defaults, if any. All that do not have a default value are required. We distinguish two types of environment variables:

• Shared: these typically have the same value across different CI/CD projects for the same Sigrid deployment. SIG recommends to configure these as variables managed by the CI/CD environment (often called “secrets”).
• Non-shared: these typically differ across projects.

Notes:

• CUSTOMER: this is the name of the Sigrid tenant as set in Sigrid’s Helm chart when Sigrid was deployed. In Sigrid, this is always a lowercase string matching regex [a-z][a-z0-9].
• SYSTEM: the name of this system (a lowercase string matching [a-z][a-z0-9-]). The default is the project name of the current CI/CD project (e.g., the pre-configured $CI_PROJECT_NAME variable in GitLab).
• SIGRID_URL: (sub-)domain where this Sigrid on-premise deployment is hosted, e.g. https://sigrid.mycompany.com.
• SIGRID_CI_TOKEN: a personal access token created in Sigrid’s UI.
• S3_ENDPOINT: URL at which an S3-compatible object store can be reached. Defaults to Amazon AWS S3 endpoints.
• BUCKET: name of the bucket in which analysis results are stored.
• AWS_ACCESS_KEY_ID: ID of the access key to authenticate to the S3-compatible object store. This key should give access to the bucket named by S3_BUCKET.
• AWS_SECRET_ACCESS_KEY: the key whose ID is AWS_ACCESS_KEY_ID.
• AWS_REGION: the region in which the bucket with name S3_BUCKET is located. For MinIO, this is us-east-1 unless a different region is configured in MinIO.
• SIGRID_SOURCES_REGISTRATION_ID: the ID of the OAuth client registration provided in values.yaml of Sigrid’s Helm chart.

Manually publishing a system to Sigrid

It is also possible to manually start an analysis, and then publish the analysis results to Sigrid. You can use this option when your system doesn’t have a pipeline, or when you need to import a system in Sigrid ad-hoc.

We recommend you integrate Sigrid CI into your pipeline. This ensures the results you see in Sigrid are always “live”, since the analysis will run after every commit. It also allows for developers to receive Sigrid feedback directly in their pull requests.

You can run the analysis and publish the analysis results using the same Docker container. If you run Sigrid CI ad-hoc, you will still need to provide the environment variables. Since there are quite some environment variables, it’s easiest to use Docker’s --env-file option for this. This option is explained in the Docker documentation.

The following example shows how to start an ad-hoc analysis for a system located in a local /mysystem directory:

docker run \
  --env-file sigrid-ci-config.txt \
  -v /mysystem:/tmp/sources \
  -ti softwareimprovementgroup/sigrid-multi-analyzer:1.0.20241206 \
  --publish

This requires you to have access to the Sigrid on-premise Docker containers.

The version tag (1.0.20241206) should match your version of Sigrid on-premise.

Contact and support

Feel free to contact SIG’s support department for any questions or issues you may have after reading this document, or when using Sigrid or Sigrid CI. Users in Europe can also contact us by phone at +31 20 314 0953.