Sigrid Axis for Sigrid On-Premise

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

Prerequisites

• You should have already read the other Sigrid On-Premise documentation, in particular Kubernetes Deployment.
• Sigrid On-Premise is installed and working, using the sigrid-stack Helm chart version 1.0.20261012 or later.

Enabling Sigrid Axis

Sigrid Axis is the product for agentic tooling: AI coding assistants use it to consult and improve your systems using Sigrid’s analysis. See the general Sigrid Axis documentation for what Axis offers and how developers connect their tools. This page covers deploying Axis in your own cluster. Axis is disabled by default in sigrid-stack.

Sigrid Axis consists of two components, which are both required:

• The Axis frontend, which runs as its own subchart in sigrid-stack.
• The Axis metrics aggregator, a cronjob that aggregates all metrics shown in Axis, including its weekly trends. Without it, the metrics in Axis are no longer updated.

The minimal configuration for both components is shown below.

global:
  onPremise:
    axisMetricsAggregator:
      image:
        tag: 1.0.20261012

sigrid-axis-frontend:
  enabled: true
  image:
    repository: softwareimprovementgroup/sigrid-axis/frontend
    tag: 1.0.20261012
  ingress:
    enabled: true
    className: "nginx"  # Use the same ingress controller as the other Sigrid services
    annotations: { }  # Specify any ingress controller-specific annotations you need
    hosts:
      - host: "axis.example.com"
        tls:
          enabled: true
    extraPaths:
      - path: /rest  # Transparently reach Sigrid's backend APIs even though the host is different
        serviceName: my-sigrid-sigrid-stack-nginx  # <release-name>-sigrid-stack-nginx

If you’re not pulling softwareimprovementgroup/sigrid-axis/frontend and softwareimprovementgroup/sigrid-axis/metrics-aggregator directly from our ECR onto your Kubernetes cluster nodes, make sure to pull them from ECR and push them to your local registry for deployment. See using SIG’s Elastic Container Registry for the registry configuration.

Sigrid Axis requires its own host name, as in the example above. It cannot be served under a path of your existing Sigrid host, because Axis serves its frontend from / of its host.

The metrics aggregator is deployed automatically when the Axis frontend is enabled. It connects to your PostgreSQL database as the same import_user used by the import jobs, reusing the secret from the import job’s PostgreSQL configuration. Each run aggregates the most recent weeks for all systems, and the current week is re-measured on every run. Results are stored in the axis_metrics schema of your sigriddb database.

Logging in to Sigrid Axis

Sigrid Axis uses the same identity provider and OpenID Connect client as Sigrid, as configured in Kubernetes Deployment. Because Axis requires its own hostname, it needs its own login callback. Register the Axis host as an additional registration in the auth-api configuration:

auth-api:
  config:
    oauth2:
      registration:
        sigridmfa:
          additionalRegistrations:
            sigrid-axis: "axis.example.com"

The label (sigrid-axis in the example) can be anything you like. Each host gets its own registration named sigridmfa- followed by the host name with dots replaced by dashes. This makes sure users return to Sigrid Axis after logging in, instead of to Sigrid Core.

You also need to allow the Axis callback URL as an additional redirect URI of the OpenID Connect client in your identity provider. The callback URL follows the pattern https://<host>/rest/auth/login/oauth2/code/sigridmfa-<host with dots replaced by dashes>. For the example above, this is https://axis.example.com/rest/auth/login/oauth2/code/sigridmfa-axis-example-com.

Metrics aggregator settings

The defaults are suitable for most deployments. The following settings can be overridden in the global.onPremise.axisMetricsAggregator section:

global:
  onPremise:
    axisMetricsAggregator:
      cronJobschedule: "0 2-5 * * *" # hourly between 02:00 and 05:00 UTC by default
      activeDeadlineSeconds: 3600 # maximum duration of a single run in seconds
      resources:
        requests:
          cpu: 1000m
          memory: 4Gi
        limits:
          memory: 4Gi

• cronJobschedule — the cron format schedule of the job. The default "0 2-5 * * *" triggers hourly during the night, which spreads the aggregation over a quiet period. A run that is skipped because the previous one is still active is picked up by the next trigger.
• activeDeadlineSeconds — how long a single run may take before Kubernetes cancels it. The default allows one hour.
• resources — the container’s CPU request and memory. The default of 4Gi memory suffices for most deployments; lower it only if your node pool cannot accommodate it.

Temporarily disabling the metrics aggregator

Should you run into issues, for example with the container or the cronjob settings, you can temporarily disable the aggregator without disabling Axis itself. This is only intended for troubleshooting; re-enable the aggregator once the issue is resolved:

global:
  onPremise:
    axisMetricsAggregator:
      enabled: false

The aggregator then stops running, and Axis will not receive updated metrics until it is enabled again. Disabling it does not remove previously aggregated data.

Contact and support

Feel free to contact SIG’s support team for any questions or issues you may have after reading this documentation or when using Sigrid.