How do I create a pipeline?

A pipeline is a named route to production for a single target state definition: environments you name yourself, in the order you want them, each holding its own domains.

Once you have created a pipeline, promoting a definition from development to test to production becomes a sequence of reviewed, approved steps you can follow on a single board.

Prerequisites

• A target state definition stored in a registered source. A pipeline promotes exactly one definition. See how to create a definition.

• Access to the domains you want to include. Domains are listed by the cell that owns them, and you only see cells you have access to.

• The entitlement to request deployments. If you can see Pipelines under Data Management but the New pipeline button is missing, you have read-only access.

Method

Step 1: Open the pipeline editor

1. Navigate to Data Management > Pipelines.

2. Select New pipeline.

The editor opens with one environment already named Development, so the draft is valid straight away. Rename it if you want.

Step 2: Name the pipeline and choose its definition

1. Enter a Pipeline name, for example Config promotion. Names must be unique among your pipelines, ignoring case.

2. In Definition source, choose the Source that holds the definition. Sources that need attention are marked with a warning icon.

3. In Target State Definition, choose the definition this pipeline promotes. The list shows each definition's scope/code and title.

4. Optionally add a Description.

Changing the source clears the chosen definition, because a definition belongs to the source it was picked from.

Step 3: Arrange the environments

Each environment is a step on the route. Configuration travels through them left to right on the board, in the order you set here.

For each environment:

1. Enter an Environment name, such as Development, UAT or Production. Every environment needs a name.

2. Pick an Environment colour from the seven swatches. The colour tints the environment's column on the board and helps you tell environments apart at a glance.

3. Optionally add a Description. It appears above the domains on the board.

4. Use Add domain to add each domain that belongs to this environment. Domains are grouped by cell. Within an environment, domains are deployed in the order listed. Use the up and down arrows beside a domain to reorder, or the cross to remove it.

To add another environment, select Add environment below the last one. Use Move environment earlier and Move environment later (the arrow buttons) to reorder, or Remove environment to delete one.

A few rules to know:

• A domain can appear in only one environment per pipeline. Domains already placed are removed from the Add domain list.

• An environment with no domains is allowed. This is how you plan a route before the domains exist. The board offers to add domains later.

• A pipeline can hold up to 20 environments and up to 100 domains in total.

Step 4: Save

Select Create pipeline. The editor opens the new pipeline's board.

Saving a pipeline never deploys anything. It only records the route.

If you leave the editor with unsaved changes, Admin Portal asks whether to Leave the editor or Stay here.

Reading the pipeline board

The board shows one column per environment, joined by arrows. Each domain appears as a node showing:

• the revision of the definition last applied there, or no revision;

• a status chip: Succeeded, Failed, Deploying now, Awaiting approval, Approved · ready, Behind latest or Never deployed;

• change counts, labelled applied for what the last run did, or expected for what the next run would do. Expected counts are not read automatically. Select Show expected changes to compute them for every waiting domain on the board in one go.

Each environment column has a Promote button when any of its domains is behind the latest revision. Promoting raises one deployment request per domain that is behind, and each request still needs approval before it can be deployed.

Linking and cloning a pipeline

Both actions appear on a pipeline's card in the list and in the board header. Both open the editor pre-filled, and nothing is saved until you select Create pipeline.

Link: share one structure across several definitions

Use Link when you want to promote a different definition along exactly the same route, and keep the two routes in step. A linked pipeline inherits its environments, colours, descriptions and domains from the original. Any change to the original's structure applies to the linked pipeline immediately.

1. On the pipeline whose structure you want to reuse, select Link.

2. The editor opens titled Linked pipeline with a read-only Inherited structure panel. Give the pipeline a name (the suggested name is the original's name followed by · linked), choose the source and the Target State Definition to promote, and add a description if you like.

3. Select Create pipeline.

On the linked pipeline you can edit only the name, description and definition. To change the environments or domains, edit the original. The board shows a banner Structure linked to [original] with a link to it.

To break the link later, select Detach on that banner. The pipeline keeps a copy of the structure as it stands and stops following the original. Detaching deploys nothing and undeploys nothing.

Clone: start from a copy you can change freely

Use Clone when you want a similar route that will diverge from the original.

1. On the pipeline you want to copy, select Clone.

2. The editor opens titled Clone pipeline with the environments, colours, descriptions, domains and definition copied in. The suggested name is the original's name followed by (copy). Change whatever you need, including the definition.

3. Select Create pipeline.

A clone is independent. Later changes to the original do not reach it, and no run history is carried over.

Editing and deleting a pipeline

• Edit pipeline on the board opens the editor. Saving replaces the whole route with what the editor shows. If other pipelines are linked to this one, the editor warns that your structure changes apply to them too.

• Delete on the board, or the bin icon on the card, removes the route. Nothing is undeployed: every domain keeps the configuration it has, and the deployments and their history stay under Deployments. Any pipelines linked to this one are detached and keep a copy of the structure.

Monitoring

The pipeline board polls the runs it shows, so status chips and progress bars advance without a refresh. For run logs and full history, select a domain's name on the board to open its deployment detail page, or use the Activity and History tabs under Data Management > Deployments.

The Pipelines list can be narrowed with the search box and the filter chips Has failures, Awaiting approval, Deploying now, Behind latest and Fully on latest. The Sort menu offers Needs attention first, which brings failed and waiting pipelines to the top.

FAQs

Can I add an approval gate or a manual pause to one environment? There is no per-environment setting, because every promotion already requires a second person to approve each request before it can be deployed. To hold configuration at an environment while you test, do not promote it further until you are ready.

Can I promote out of order? Yes. If an earlier environment is not on the latest revision, the Promote button for a later one is de-emphasised and its tooltip says the promotion skips a step, but it is still allowed. Every request is planned and approved on its own.

Can one definition have several pipelines? Yes. There is no limit. Link and Clone are the quickest way to make a second route for the same or a different definition.

Can a pipeline mix domains from different cells? The editor lets you pick domains from any cell you can see. See the outstanding questions below.

What happens to runs in progress if I delete the pipeline? They continue. Their history stays under Deployments, and the domain's Deployed via column reads Deployed directly once the pipeline is gone.

Appendix: Limits