Programmable API
A REST API on Wave Core and 14 Custom Resource kinds, so a script or a Git repository can change Wave settings instead of the console.
Turning on Autopilot for 40 Deployments means 40 trips through the console
Wave Architecture docs; Wave CRD Mode docs; Wave API reference (v3). The 40 Deployments are illustrative.
The console changes one workload at a time. To turn on Autopilot for 40 Deployments, someone opens each one, changes the setting, and saves it, 40 times.
Wave Core already serves a REST API on port 3024, "for Web Console and external integrations" (Wave Architecture docs). A script can make the same calls the console makes.
In CRD Mode, the settings live in Git as Custom Resources, so the same YAML applies to every cluster (Wave CRD Mode docs).
Endpoint: Wave API reference (v3). REST API on Wave Core, port 3024: Wave Architecture docs. The 40 Deployments, their names, and the script are illustrative.
How Wave does it
- Wave Consolepick the cluster, open the list
- openDeploymentfind it by name, open its page
- editAutopilotturn it on in a dialog, then save
- repeatNext oneback to the list, 39 to go
- Your scripta CI job or a shell loop
- PUTWave CoreREST API on port 3024
- appliesDeploymentAutopilot on, as the console would set it
- nextAll 40changed in one run, no dialog
Call the REST API
The Wave API reference (v3) lists each endpoint with its parameters and responses. For example, one PUT call to the autopilot-enabled endpoint of a Deployment turns Autopilot on or off for that Deployment.
Or keep the settings in Git
In CRD Mode, 14 kinds of wavek8s.com Custom Resources hold the settings. You apply them with kubectl or ArgoCD, and Wave applies the same settings the console would. The console becomes read-only for those settings.
Check that it applied
Each Custom Resource reports a Synced condition in .status, with a reason such as AsExpected, TargetNotFound, or ValidationFailed. A target that does not exist yet is applied when it appears.