Pipelines
A pipeline deploys several components of an environment in ordered stages, in parallel within a stage, stopping or continuing when a stage fails.
A pipeline deploys several components of one environment in a set order: for example, run the database migration job, then deploy the API and the worker together, then the web frontend.
Stages
A pipeline is a list of stages, shown as steps in the console, run one after the other. Each stage lists the components it deploys, and all components of a stage deploy at the same time. The next stage starts once every component of the current one has finished. A component appears in at most one stage of a pipeline, so it deploys once per run.
Each component in a stage has two settings. The console adds components as required with no timeout; the Kapten API can change both.
| Setting | Description |
|---|---|
| Required | When a required component fails, its stage fails. A component that is not required can fail without failing the stage. |
| Timeout | How long, in seconds, the component’s deployment may take before it counts as failed. |
When a stage fails
Each stage has a failure policy:
| Policy | Behavior |
|---|---|
Stop the pipeline (STOP_PIPELINE) |
The pipeline stops. The stages after it are skipped. |
Continue with the next step (CONTINUE_PIPELINE) |
The pipeline goes on with the next stage, and is reported as failed at the end. |
Runs
Starting a pipeline creates a pipeline run. It records the status and timing of the pipeline, of each stage and of each component deployment, with a message explaining any failure. Each component deployment is also a run of the component, with its own logs.
A pipeline can be deactivated to keep its definition without running it.