Orbit
Orbit Deployment Pipeline
The Pipeline tab is a single page that answers "what happens when we push". It shows every way a deploy can be triggered, every gate that can stop one, the state of each environment, and which build features are switched on, all read from your project's real configuration.
Where the Pipeline Lives
Open Orbit, click the project, and choose Pipeline under the Deployments group in the project tab strip. The page is labelled Deployment pipeline.
It is a read-only dashboard. Nothing is configured here; every section links through to wherever the setting actually lives. That is its value: one screen to understand the project, rather than reading eight settings cards.
Alerts At the Top
Two banners appear when they apply:
- Deploy lock active, with a Manage link. Push-triggered deploys are being skipped.
- N deployments awaiting approval, with a Review link. Someone needs to approve or reject them.
If either is showing and you are wondering why a push has not deployed, you have your answer without reading further.
Build Stats
A compact panel shows, over the recent builds: success rate, average build time, and how many succeeded. It is a health check rather than an analysis. For the full picture, see Orbit Project Analytics and Orbit Build Insights.
Trigger Sources
This section lists every route into a deploy for this project:
| Source | What it shows |
|---|---|
| Git push | The connected repository, or No repo connected |
| Branch previews | Whether previews are created automatically on any branch |
| Git tag | The tag pattern, if one is configured |
| Manual deploy | Always available |
Read this whenever you are surprised by a deployment. If a build appeared and nobody pushed, one of these is the explanation: a tag, a deploy hook, or someone pressing a button.
Gates and Safety
The largest section is the list of gates, each showing its current state with a Configure link to the relevant settings card.
| Gate | What it does |
|---|---|
| Approval gate | Production deploys require explicit approval |
| Staging prerequisite | Production waits for staging at the same commit |
| CI checks | Required checks that must pass, or No checks required |
| Freeze window | Weekends blocked, custom hours, or no freeze schedule |
| Deploy lock | Active, or no lock |
| Health check | The path checked after deploy, or Disabled |
| Auto-rollback | Reverts on health check failure |
| Skew protection | The retention window for old assets |
| Auto retry | How many times infrastructure failures are retried |
Most of these gates apply only to push-triggered deploys. Manual deploys from the panel and deploy hooks pass straight through. The exception behaves differently, and the detail of each gate is covered in Deploying Your Project. Read that before you rely on a gate as a control.
Reading the Gates As a Checklist
For a project that matters, a sensible baseline is:
- Health check: set, pointing at a path that exercises the app rather than a cached shell.
- Auto-rollback: on. Without a health check it has nothing to act on, so the two go together.
- Auto retry: one or two. It re-queues builds that failed on infrastructure errors such as a network blip, and does not retry code errors, so it costs you nothing but time saved.
- Approval gate: on for anything where a bad deploy is expensive, off where it slows you down more than it protects you.
If the page shows Health check Disabled and Auto-rollback on, that combination does nothing. It is one of the easiest misconfigurations to carry for months without noticing, and this page is where you spot it.
Environment Flow
The Environment flow section draws each environment as a card with its current state: LIVE, BUILDING or PAUSED, the branch it tracks, and the retry number if the current deployment is a retry.
Badges on each card show what is enabled for that environment:
- Smoke tests, GET requests run against chosen paths after each successful deploy.
- Auto-promote, promoting staging to production after a number of healthy hours.
- Canary, the percentage of traffic on a canary deployment.
- Inherits prod vars, where staging merges production environment variables at lower priority.
- Scheduled rebuild, where production rebuilds on an interval.
Each card links to that environment's deployments. If an environment says No deployments yet, it exists in configuration but nothing has shipped to it.
Active Features
The last section summarises build-level settings:
| Feature | Values |
|---|---|
| Server mode | SSR enabled or static only |
| Auto-create on push | Whether branch pushes create environments |
| Health checks | On or off |
| Build retry | A maximum, or off |
| Deploy groups | Grouped with other projects, or standalone |
| Build timeout | The per-build limit |
| Preview expiry | Days before previews are paused, or never |
Server mode is the one people get caught by. A framework that renders on the server needs it on; a static export does not. If your project builds fine and then serves a blank page or a 404 on every route but the home page, check this first. See Frameworks Orbit Supports.
Preview expiry is the housekeeping one. Set to never, preview environments accumulate indefinitely.
Using the Pipeline Page
When onboarding someone. Send them here first. It is a faster and more accurate briefing than any document, because it is generated from the live configuration.
When a deploy did not happen. Work top down: banners, then trigger sources, then gates. One of the three will explain it.
Before a risky release. Check the gates section reads the way you think it does. It is the difference between believing you have auto-rollback and having it.
During an incident. The environment flow tells you what is live where, and whether anything is mid-build.
Troubleshooting
The page shows No repo connected. The project has no repository. Connect one: see Connecting a GitHub Repository.
A gate is on but deploys still go through. It is push-triggered only. A deploy hook or a manual panel deploy is not affected.
An environment shows PAUSED. Preview environments pause automatically once past their expiry. Redeploy to bring one back.
Auto-promote is shown but nothing promotes. It requires the configured number of healthy hours with smoke tests passing, and is evaluated periodically rather than instantly.
Where To Go Next
- Deploying Your Project for what each gate actually blocks.
- Orbit Project Settings to change any of it.
- Rolling Back a Deployment when a gate did not save you.