This release focuses on significant improvements to the deployment process, removal of deprecated components, and enhanced stability. Key highlights include the replacement of the install job with apl-operator, removal of several deprecated monitoring and maintenance tools, and numerous fixes to improve upgrade reliability and ArgoCD operations.

📘
For a full list of changes, review the release notes on GitHub: Release v4.13.0.

New

Installation now relies on a reconciliation loop of the apl-operator to improve overall robustness. Previously, the Kubernetes job was responsible for the initial installation.
ArgoCD Application sources are leveraged to fetch Helm chart values from two files that are stored in the values repository.
Added endpointslice availability check for oauth2proxy to improve Kubernetes compatibility
CoreDNS is now used for LKE Enterprise clusters to improve DNS resolution
Platform database connection pooling settings are now user configurable.

Changed

Updated core applications. See the Core application changes table below for details.

Fixed

Minor improvements to runtime upgrades
Resolved Helm issue that was blocking the installation retry mechanism
Fixed Kubeflow Pipelines (KFP) workflow image references
Corrected service names in OpenTelemetry CRDs
Force apply essential manifests on upgrade to prevent deployment failures
Fixed installation failures when initializing local charts repository by adding retry logic
Fixed string quoting in Loki config to avoid unintended type conversion
Addressed PostgreSQL image upgrade considerations for 4.12 release compatibility
Fixed bash launch on TTY startup and embedded all required changes in TTY image
Fixed failing jwt verification
Fixed missing status field for APL resources
Fixed config for express rate limit
Fixed values editor required secret fields
Fixed team privilege escalation through API
Fixed improper ACL on team metadata

Deprecated

Deprecated the Tempo app.
Removed the Falco, HTTPbin, Jaeger, Kiali, Kured, and Velero apps.

Core application changes

App name	Old version	New version	Notes
Alertmanager	0.27.0	0.29.0	Updated
Argo CD	3.1.8	3.2.0	Updated
Cert-manager	1.19.0	1.19.1	Updated
CloudNative PostgreSQL Operator	1.27.0	1.27.1	Updated
Falco	0.36.2	-	Removed
Gitea	1.24.5	1.24.7	Updated
Grafana	12.0.2	12.2.1	Updated
Harbor	2.13.2	2.14.0	Updated
HTTPbin	0.1.0		Removed
Ingress-NGINX	1.13.3	1.14.0	Updated
Istio	1.26.3	1.27.3	Updated
Jaeger	1.46.0		Removed
Keycloak	26.3.3	26.4.2	Updated
Kiali	2.10.0		Removed
Kserve	0.15.2	0.16.0	Updated
Kyverno	1.15.1	1.15.2	Updated
Kured	1.13.1		Removed
Loki	2.9.10	2.9.15	Updated
Prometheus	3.4.2	3.7.3	Updated
Sealed Secrets	0.32.2	0.33.1	Updated
Tempo	2.6.0		Deprecated
Trivy Operator	0.28.0	0.29.0	Updated
Open Telemetry Operator	0.80.0	0.138.0	Updated
Velero	1.9.0		Removed
Policy Reporter	3.4.2	3.5.0	Updated

Post-upgrade actions

After upgrading to v4.13.0, you may need to perform a few additional steps depending on the features and apps that are used on your App Platform deployment.

If you are using the Auto Image Updater feature for any workloads, there are some changes to file locations within your local Gitea repositories. See Change repository when using Auto Image Updater for a workload .
If you are using any of the apps that have been removed in this release and want to remove and clean up their resources, see Clean up resources for apps that have been removed .

Change repository when using Auto Image Updater for a workload

Users that are leveraging a workload’s Auto Image Updater feature may need to transfer a file between Gitea repositories. For each workload, move and rename the original file below from the otomi/charts Gitea repository to a new file in the otomi/values Gitea repository.

Old file (in otomi/charts): <chartName>/.argocd-source-<teamId>-<workloadName>.yaml
New file (in otomi/values ): env/teams/<teamId>/workloadValues/<workloadName>.managed.yaml

Clean up resources for apps that have been removed

A few deprecated apps have been removed in this release and are no longer supported. App Platform doesn’t automatically clean up all resources for these apps as some users may continue using them until they are able to transition to other apps. If you do wish to remove these apps and their associated resources, use the commands below.

Falco

kubectl delete application falco-falco -n argocd
kubectl delete application falco-falco-exporter -n argocd
kubectl delete namespace falco

HTTPbin

kubectl delete namespace httpbin

Jaeger

kubectl delete application jaeger-operator-jaeger-operator -n argocd
kubectl delete namespace jaeger
kubectl delete namespace jaeger-operator
kubectl get crd | grep -i jaeger (finds all jaeger crds)
kubectl delete crd jaegers.jaegertracing.io

Kiali

kubectl delete application kiali-operator-kiali-operator -n argocd
kubectl delete application kiali-kiali-operator-artifacts -n argocd
kubectl delete namespace kiali
kubectl delete namespace kiali-operator
kubectl get crd | grep -i kiali (finds all Kiali crds)
kubectl delete crd kialis.kiali.io

Kured\

🚧
Be careful when removing resources for kured as some live in the kube-system namespace. To prevent unintended deletions, delete resources in that namespace by label.

kubectl delete application kube-system-kured -n argocd
kubectl delete all -n kube-system -l app.kubernetes.io/name=kured

Velero

kubectl delete namespace velero

Changelog