esc

Docs

Releases and rollback

What happens when a release is approved, how health checks work, and how to roll back.

What happens on approve

When a production approval is approved, DeployNX:

  1. creates the release tag on the approved commit;
  2. runs the production deploy pipeline you confirmed for the repository;
  3. calls the production health URL and records the status code and response time;
  4. marks the release healthy, or unhealthy if the health check still fails after its attempts (5 by default), and rolls an unhealthy Stage or Production release back to the last healthy one.

Every step lands on the release’s timeline, with who requested it and who approved it.

Owners and admins can change these for the organization in Settings › General: how many times the health check is tried, whether unhealthy releases roll back automatically, and how long a deploy pipeline is followed before it’s marked timed out (30 minutes by default).

Guard rails

  • One deploy at a time per repository and environment. DeployNX won’t start a second deploy while one is running.
  • Healthy environment first. Production waits until the environment before it on the repository’s route is healthy (Stage, unless the repository maps other environments before Production).
  • GitHub environments. If your production job uses a GitHub environment: with required reviewers, DeployNX releases that gate as the approver, so nobody approves twice. The approver must be a required reviewer on that environment and must have authorized the DeployNX GitHub App.

Environments

Every organization starts with Stage and Production. Add your own in Settings › Environments, such as QA, UAT or a regional production, and set each one’s tier, its place in the promotion order and who approves releases to it. To add the environments a repository’s CI config deploys to, choose Add detected environments in its settings; DeployNX maps its pipelines to them. A repository releases along the environments it maps, in your organization’s order; one that maps none, or that auto-detects green stage builds, uses Stage, then Production. An environment with releases can’t be removed. Any environment other than Stage and Production can be archived instead, which takes it off every route until you restore it.

Environments shows your organization’s environments for every repository: the tag that’s live, when it shipped, its health and its response time. When an environment runs an older release than the one before it, a Promote button opens the promote flow.

Rolling back

Open any shipped release and choose Roll back. Pick the release to return to, give a reason, and confirm. DeployNX runs your production deploy for that tag and checks health again. The rollback, its reason and who asked for it are recorded on the release timeline and in the audit log.

Make DeployNX the only path to production

DeployNX can only record releases that go through it. To stop deploys that skip it, limit who can run your production pipeline directly on your git host: protected environments or deployment permissions on GitHub, Bitbucket and GitLab, with DeployNX’s connection as the one allowed actor.

NextChat, webhooks and API tokens