Skip to main content

Workflows

A workflow is the lifecycle of an install or a component. It plans the change, surfaces a diff, runs through any approvals you’ve configured. Each workflow run’s logs and step state are visible in the dashboard, CLI, and TUI. How it works → Workflows

Health Checks

Nuon checks the health of every component on every install, with no agent to install. Helm chart and Kubernetes manifest components are checked about once a minute. Terraform components get a health status when you add an HTTP or TCP probe, or a custom check. When a component or an install becomes unhealthy, and when it recovers, Nuon sends an alert to Slack or a webhook. Each component and install keeps 90 days of timeline and uptime. How to → Health Checks

Install Telemetry

Install telemetry is the logs, metrics, and traces your application produces while running in a customer’s cloud account. Nuon can forward this data to your observability backend so you can monitor performance and troubleshoot issues across installs. Forwarding is off until you set it up, and it needs a Nuon BYOC deployment. Without it, Nuon stores only the logs of its own operations and the output of your actions. Application logs stay in the customer’s account. How it works → Install Telemetry

Audit Logs

Install runners on AWS, Azure, and GCP can forward deployment, action workflow, and sandbox audit events directly to a customer-controlled OTLP logging backend. The destination and credentials remain in the customer’s cloud account. Only Nuon audit events are forwarded, not application logs or general runner logs. How to → Export Runner Audit Logs

Actions

An action is a script you run on a live install. The runner executes it inside the customer’s account, so you need no SSH or kubectl access to the customer’s network. Define it in TOML, point it at a repo, and set environment variables. It runs on a cron schedule, on a lifecycle event such as post-deploy-component or pre-reprovision, or on a manual trigger. The output and run history stay in Nuon. During an incident, run a one-off command with an adhoc action without changing your app config. Chain actions and deploys into a repeatable procedure with a runbook. Typical uses: database migrations, maintenance scripts, collecting logs, and custom checks that the built-in health checks do not cover. How it works → Actions

Database Migrations

There are three ways to run migrations across installs: a Kubernetes Job that an action starts after each deploy, an init container in your app’s pods, or an action that runs a container image on the install runner. Use an action with a pre-deploy-component trigger when the migration must finish before new pods start. A failed migration action pauses the deploy until someone retries or skips it. How to → Database Migrations

Policies

A policy is an OPA or Kyverno rule that gates what can be deployed. Policies validate Terraform plans, Helm charts, Kubernetes manifests, container images, and sandbox configurations before they apply. Every evaluation is logged on a per-app analytics page, scoped to the install, component, action, or deploy that triggered it. Use policies to enforce compliance, security baselines, or organization conventions across every customer install at once, instead of relying on review discipline per install. How it works → Policies

Operation Roles

An operation role is a specific IAM role the runner assumes for one class of work: provisioning, deploys, or individual actions. For urgent elevated access, use break glass roles. Each role is defined by the vendor and approved by the customer through the Stack, so the runner never holds more permissions than the task at hand requires. Roles can be scoped per component or per action; every workflow run records which role was used. How it works → Operation Roles