Parallel releases
Parallel releases run multiple non-conflicting selective releases at once in the same environment, across every resource type, with queuing or state-lock handling.
Parallel releases run multiple Selective Releases in the same environment at once. Every resource type is covered, so a release that touches many independent resources finishes faster than processing them one at a time. Non-conflicting releases run in parallel; conflicting ones are queued or fail with a state-lock error.
A release parked for approval blocks parallel execution. See Approval workflow.
Enable parallel releases
- Navigate to Settings > General to view the Control Plane settings.
- Set Enable Parallel Releases to true (this is set to false by default).
- Specify the Maximum Parallel Releases that can be executed across the Control Plane.
Note: The default value for Maximum Parallel Releases is set to 10. - Click Save Changes.
Use parallel releases
Triggering releases
Perform multiple Selective Releases as usual. The Control Plane executes non-conflicting releases in parallel on its own, with nothing extra to set at trigger time.
Using prompt
Release service/api and postgres/main-db in dev for the checkout project.or
Using command
raptor create release -p PROJECT -e ENVIRONMENT --target service/api --target postgres/main-dbRepeat --target for each resource, written as TYPE/NAME. Add --no-queue to fail straight away when another release is already running in the environment, instead of joining the queue.
New to the CLI? Install it first.
Follow the selective release steps, choosing Selective under Release type and picking the resources you want to deploy.
Execution logic
- Non-conflicting releases run in parallel.
- Conflicting releases are queued or fail with a state lock error, requiring manual re-triggering.
Conflicts and error handling
- Parallel execution supports Selective Updates and Selective Create/Delete unless they target the same resource.
- Conflicting releases involving shared resources or Selective Create/Delete result in queuing or state lock errors.
- Re-trigger failed releases after resolving conflicts for smooth deployment.
A queued release has no release id until it starts. raptor get queued-releases -p PROJECT -e ENVIRONMENT lists what is waiting, and the trace id it reports follows one into the release it becomes, via raptor get releases --trace-id TRACE_ID.
Rollback support
- Rollbacks are supported, but a release with refresh enabled may be necessary to fully restore the system.
Canary releases
Configure Argo Rollouts canary deployments for Facets services: set rollout behavior, trigger a release, monitor pods, then promote or abort.
Guardrail policies
Guardrail policies in Facets enforce security and compliance on blueprints using Rego, with Warning or Error severity that can gate releases.