RATATOSK
← All releases

Knativeknative-v1.20.1

Security 1Breaking changes 2Important fixes 3Dependency updates 1

Get Knative alerts

Email is sent weekly only when there are changes to check; critical security advisories are sent immediately.

Security1

AllowRootBounded security-context defaults

AffectsRevision workloads using the secure-pod-defaults feature setting(Inferred)
Additional actionsecure-pod-defaults set to AllowRootBounded

AllowRootBounded defaults SeccompProfile and Capabilities when they are nil, without applying the runAsNonRoot default used by Enabled. In v1.20, secure-pod-defaults remains disabled by default, so upgrading alone does not change Revision defaults.

Show release-note lines
We've introduce `secure-pod-defaults` in an earlier release but this release includes a new setting `AllowRootBounded` that offers a better security posture for your workloads but balances the compatibility with images that require/expect you to run as root.
github.com/knative/serving/releases/tag/knative-v1.20.1

Breaking changes2

Serving metric names updated

AffectsServing metric queries or dashboards using kn.queueproxy.app.duration or kn.queueproxy.depth(Inferred)

Serving now emits kn.serving.invocation.duration and kn.serving.queue.depth instead of kn.queueproxy.app.duration and kn.queueproxy.depth.

Evidence“Fix metric names to match the original design document `kn.queueproxy.app.duration` becomes `kn.serving.invocation.duration` and `kn.queueproxy.depth` becomes `kn.serving.queue.depth`” knative/serving/pull/16293 ↗
Show release-note lines
* [release-1.20] Fix Serving Metric Names to match Design Document by @knative-prow-robot in https://github.com/knative/serving/pull/16293
github.com/knative/serving/releases/tag/knative-v1.20.1

Autoscaler service stats metrics drop RevisionNameKey

AffectsSetups consuming autoscaler service stats metrics that use RevisionNameKey(Inferred)

Before this release, the scraper attached the revision name to these metrics as RevisionNameKey.

Show release-note lines
* [release-1.20] Make autoscaler instrument async so we can remove metric attributes when revisions go away by @knative-prow-robot in https://github.com/knative/serving/pull/16302
github.com/knative/serving/releases/tag/knative-v1.20.1

Important fixes3

  • Knative Serving preserves external metadata during reconciliation
    AffectsServing Deployments or pod templates with external labels or annotations(Inferred)

    Knative-controlled knative.dev keys and the app label use desired values; other metadata keys are preserved from the current resources.

    * [release-1.20] Preserve deployment and its template labels and annotations by @knative-prow-robot in https://github.com/knative/serving/pull/16303
    github.com/knative/serving/releases/tag/knative-v1.20.1
  • Unbounded autoscaler metric growth fixed
    AffectsKnative Serving autoscaler metrics collection(Inferred)

    When revision metric attributes could not be removed after revisions went away, metric data could grow without bound. Autoscaler metrics now use observable gauges, and their registrations can be unregistered when the related scaler resources are deleted.

    * [release-1.20] Make autoscaler instrument async so we can remove metric attributes when revisions go away by @knative-prow-robot in https://github.com/knative/serving/pull/16302
    github.com/knative/serving/releases/tag/knative-v1.20.1
  • Serving now suppresses the “default value insecure” warning
    AffectsServing setups with SecurePodDefaults set to Enabled or AllowRootBounded

    This includes the tested incomplete security-context case.

    * [release-1.20] Suppress 'default value insecure' warning when secure-pod-defaults is enabled. by @knative-prow-robot in https://github.com/knative/serving/pull/16220
    github.com/knative/serving/releases/tag/knative-v1.20.1

Dependency updates1

    1 more, none mentioning security
    • * [release-1.20] Upgrade to latest dependencies by @knative-automation in https://github.com/knative/serving/pull/16319

    Summaries are produced by AI, so they can be wrong, and they are provided without warranty.