Workspace Runtime Lifecycle

Every Crawbl user gets an isolated Agent Runtime pod, provisioned through a Kubernetes custom resource. The platform operator watches these resources and reconciles the full runtime environment: workload, storage, services, and configuration.

State Machine

A workspace runtime resource progresses through phases tracked in status.phase.

Click diagram to zoom

1

Step 1

Progressing

The operator is converging child resources (pod identity, configuration, storage, services, or workload).

2

Step 2

Ready

The workload has at least one ready replica and the Ready condition is true.

3

Step 3

Suspended

Workload scaled to zero replicas. Storage and services remain for later resume.

4

Step 4

Deleting

The operator is removing child resources and waiting until all are gone.

5

Step 5

Error

The operator could not build the desired runtime state, usually due to invalid bootstrap or runtime config.

The backend derives a user-facing runtime.status from the phase:

• ready -- runtime is verified
• provisioning -- Progressing, Deleting, or empty phase
• offline -- Suspended
• failed -- Error

What Gets Created

For each workspace, the operator provisions identity, configuration, storage, networking, and the runtime workload from a single high-level resource.

Suspend and Resume

Setting the suspend flag scales the workload to zero. Storage and services remain. Clearing the flag resumes without losing state.

Cleanup

When a workspace runtime resource is deleted, the operator removes all child resources in the shared runtime namespace and waits until all are gone before completing deletion.

What's Next

See the Agent Runtime to learn what runs inside the agent pod.