Access Probes¶

An access startup probe verifies that the workspace’s routing resources are serving traffic before the controller marks the workspace as Available. A WorkspaceAccessStrategy configures it to act as a one-shot gate — once the probe passes, it is never re-checked until the workspace restarts or the access strategy spec changes.

Configuration¶

An access strategy configures its probe in the spec.accessStartupProbe attribute:

apiVersion: workspace.jupyter.org/v1alpha1
kind: WorkspaceAccessStrategy
metadata:
  name: my-strategy
spec:
  accessStartupProbe:
    httpGet:
      urlTemplate: "https://{{ .Service.Name }}.{{ .Service.Namespace }}.svc.cluster.local:443/healthz"
    initialDelaySeconds: 5
    periodSeconds: 2
    timeoutSeconds: 5
    failureThreshold: 30

Parameters¶

Field	Default	Description
`initialDelaySeconds`	`0`	Seconds after access resources are created before probing starts
`periodSeconds`	`2`	How often to perform the probe
`timeoutSeconds`	`5`	Seconds before the probe times out
`failureThreshold`	`30`	Consecutive failures before marking the workspace `Degraded`

URL template¶

The urlTemplate is a Go text/template with access to the same variables available in access resource templates:

.Workspace — the Workspace object
.AccessStrategy — the WorkspaceAccessStrategy object
.Service — the workspace’s Service

Behavior¶

After the workspace’s access resources are created and initialDelaySeconds has elapsed, the controller begins probing.
An HTTP GET is sent to the resolved URL. Status codes 200–399 are considered success (additional codes can be allowed via additionalSuccessStatusCodes).
On the first success, status.accessStartupProbeSucceeded is set to true and the workspace transitions to Available.
If failures reach failureThreshold, the workspace is marked Degraded. It must be stopped and restarted to retry.

Probe reset¶

The probe state resets when:

The workspace is stopped and restarted.
The access strategy specs change (detected via status.observedAccessStrategyVersion).