OpenShift docs are moving and will soon only be available at docs.redhat.com, the home of all Red Hat product documentation. Explore the new docs experience today.
  1. Documentation
    2. history bug_report picture_as_pdf
×

Startup probes

Startup probes verify whether a service has started successfully, helping to reduce cold start times for containers with slow startup processes. Startup probes run only during the container’s initialization phase and do not execute periodically. If a startup probe fails, the container adheres to the defined restartPolicy.

Progress deadline

By default, services have a progress deadline that defines the time limit for a service to complete its initial startup. When using startup probes, ensure that the progress deadline is set to exceed the maximum time required by the startup probes. If the progress deadline is set too low, the startup probes might not finish before the deadline is reached, which can prevent the service from starting.

Consider increasing the progress deadline if you encounter any of these conditions in your deployment:

  • The service image takes a long time to pull due to its size.

  • The service takes a long time to become READY because of initial cache priming.

  • The cluster relies on autoscaling to allocate resources for new pods.

Configuring startup probing

For OpenShift Serverless Serving, startup probes are not defined by default. You can define startup probes for your containers in your deployment configuration.

Procedure

  • Define startup probes for your service by modifying your deployment configuration. The following example shows a configuration with two containers:

    Example of defined starup probes
    apiVersion: serving.knative.dev/v1
kind: Service
# ...
spec:
  template:
    spec:
       containers:
        - name: first-container
          image: <image>
          ports:
            - containerPort: 8080
          # ...
          startupProbe: (1)
          httpGet:
            port: 8080
            path: "/"
        - name: second-container
          image: <image>
          # ...
          startupProbe: (2)
          httpGet:
            port: 8081
            path: "/"
    1 Startup probe of the first-container.
    2 Startup probe of the second-container.

Configuring the progress deadline

You can configure progress deadline settings to specify the maximum time allowed for your deployment to progress before the system reports a failure for the Knative Revision. This time limit can be specified in seconds or minutes.

To configure the progress deadline effectively, consider the following parameters:

  • initialDelaySeconds

  • failureThreshold

  • periodSeconds

  • timeoutSeconds

If the initial scale is not achieved within the specified time limit, the Knative Autoscaler component scales the revision to 0, and the Knative service enters a terminal Failed state.

By default, the progress deadline is set to 600 seconds. This value is specified as a Golang time.Duration string and must be rounded to the nearest second.

Procedure

  • To configure the progress deadline setting, use an annotation in your deployment configuration.

    Example of progress deadline set to 60 seconds
    apiVersion: serving.knative.dev/v1
kind: Service
...
spec:
  template:
    metadata:
       annotations:
            serving.knative.dev/progress-deadline: "60s"
    spec:
        containers:
            - image: ghcr.io/knative/helloworld-go:latest