Release date: July 15, 2026
If you're running Dynatrace Operator version 1.10.0, we recommend upgrading to version 1.10.2 to receive the latest important patches.
On this page, you'll find an overview of what's new and improved in Dynatrace Operator version 1.10.0.
Dynatrace now provides a central place to manage metadata enrichment across all telemetry data and data sources in Kubernetes. Through this central configuration, you can:
For more information, see Enrich Kubernetes telemetry with primary Grail fields and tags.
Dynatrace now supports auto-update for Dynatrace images1 hosted on supported public registries, bringing the same seamless update experience previously available through the Dynatrace built-in registry.
Supported public registries offer multi-arch images compatible with ARM64 (AArch64), x86-64, s390x, and PPC64le architectures on Linux, adhering to best practices for immutability and signing to strengthen supply chain security.
For configuration details and more, see Use a public registry.
Dynatrace Operator automatically resolves and updates container images for ActiveGate, OneAgent, OneAgent Log Module, Code Modules, Extension Execution Controller (EEC), and SQL Extension Executor.
Dynatrace Operator now supports platform tokens, the successor to access tokens that offer a more secure and unified way to authenticate with the Dynatrace platform. No immediate action is required and existing access tokens continue to be accepted. We will inform you with appropriate lead time when a migration to platform tokens becomes mandatory for Dynatrace Operator installations.
For more details about platform tokens and how to adopt them, see Tokens and permissions.
When a platform token is provided, Dynatrace Operator no longer sends mark-for-termination events. In Kubernetes environments — especially those using cluster auto-scalers — it is difficult to reliably distinguish an intentional node removal from an unexpected node failure. This ambiguity caused false-positive "Host is unavailable" alerts and reduced alerting accuracy. Host availability events on host shutdown and reboot, introduced in OneAgent version 1.301, now provide more reliable termination signals from inside the host, superseding the mark-for-termination event.
cloudNativeFullstack, applicationMonitoring, and hostMonitoring modes. For details, see Declarative process grouping.csidriver.migrationMode Helm value lets you enable this feature. While migration mode is active, the CSI DaemonSet continues running so existing CSI mounts can be cleanly unmounted, and all newly injected pods immediately use ephemeral volumes. For step-by-step instructions, see Migrate from CSI driver to ephemeral volumes.troubleshoot command now reports recently OOMKilled ActiveGate pods in troubleshoot.txt, making it easier to identify out-of-memory issues via the support archive.webhook-cert-generator and crd-storage-migrator), making startup more deterministic and easier to diagnose. webhook-cert-generator is skipped on OpenShift and OLM marketplace installations, where the platform manages certificates. The crd-storage-migrator init container is controlled by the operator.crdStorageMigrationInitManager Helm value (enabled by default for new installations).securityContext field on deployed components. Dynatrace Operator version 1.10.0 extends this by also translating the deprecated AppArmor annotation on Pod templates into the corresponding container security context (Kubernetes 1.31+) and stripping the legacy annotation. This applies to all deployed components and is a temporary measure until a dedicated field in the DynaKube and EdgeConnect spec is available. No action is required. For details, see Enable AppArmor for enhanced security.image.digest Helm value, separately from image.tag. When image.digest is set, it is used to pull the image, while image.tag continues to be used as the version label. This matches the new imageRef.digest field on the DynaKube spec and follows the Kubernetes recommendation to deploy workloads by digest.feature.dynatrace.com/node-image-pull feature flag now only affects the CSI driver. CSI-less codeModulesImage injection does not require this flag and is not affected by it.spec.telemetryIngest is configured and spec.metaDataEnrichment is enabled, Dynatrace Operator now automatically enriches all ingested telemetry with the full set of resource attributes.whenUnsatisfiable policy for the ActiveGate TopologySpreadConstraint has changed from DoNotSchedule to ScheduleAnyway. This makes ActiveGate updates more reliable on clusters with few nodes, where DoNotSchedule could prevent pods from being scheduled during rolling updates. This change causes an ActiveGate restart on upgrade.ClusterRole manifests are now available. These manifests grant exactly the permissions needed to install the Operator, with variants for clusters with and without the CSI driver, and for environments that prohibit the escalate and bind RBAC verbs. For details, see Deployment permissions.Since Dynatrace Operator 1.9.0, a RuntimeDefault seccomp profile is applied to the Dynatrace init container by default. On OpenShift, this can interfere with SecurityContextConstraints (SCCs) — such as anyuid, restricted, or nonroot — that prevent seccomp profile usage, causing the system to fall back to a different SCC (for example restricted-v2). This may render application pods unschedulable or cause workload degradation.
networkZone is configured on the DynaKube, Dynatrace Operator now waits for the in-cluster routing ActiveGate to register with the Dynatrace environment before deploying OneAgent. This prevents connection failures during initial setup of restricted network-zone deployments.nodes/proxy permission is not allowed. The dynatrace-logmonitoring ClusterRole now includes the fine-grained nodes/pods permission when deploying to Kubernetes 1.36 or later, where Fine-Grained Kubelet API Authorization is now GA. You can also set the Helm value rbac.kubeletFineGrainedAuthz to true to enable nodes/pods on older Kubernetes versions, or to false to enforce nodes/proxy on newer clusters.dynatrace-operator-crd-storage-migration job failed with BackoffLimitExceeded due to expired webhook certificates. The certificate-generation init container now runs before CRD storage migration, ensuring certificates are renewed first. This applies only when the crdStorageMigrationJob Helm value is enabled (disabled by default).otelCollector is configured but telemetryIngest is not enabled.CrashLoopBackOff when webhook.hostNetwork was enabled — a setting required by some CNI plugins such as Calico on AWS EKS. The startup probe's DNS lookup for kubernetes.default.svc failed in host network mode without the correct DNS policy. Dynatrace Operator now automatically sets dnsPolicy: ClusterFirstWithHostNet when webhook.hostNetwork is enabled.spec.templates.otelCollector.annotations were not consistently propagated to the OpenTelemetry collector.dtclient.config field of the bootstrapper-config secret and is now passed to the bootstrapper init container as a command-line argument.RegisterPlugin failure, leaving the CSI driver unregistered on the node. The registrar container is now restarted to repeat the registration procedure.extractCodeModulesImageLinks Helm value to true, but this option will be removed in a future release.logDir and dataStorageDir were missing from ruxitagentproc.conf.resources did not match the main container’s and could not be configured via the oneAgentResources field.Deprecated k8s-app-enabled feature flag: The feature.dynatrace.com/k8s-app-enabled feature flag is deprecated. The builtin:app-transition.kubernetes settings schema it previously created is no longer available on newer Dynatrace environments, where the Kubernetes app experience is enabled automatically. Setting this flag now triggers a validation warning. Remove this annotation from your DynaKube configuration. For details, see DynaKube feature flags.
The gke-autopilot.yaml file has been removed from GitHub release artifacts. Use Helm to deploy Dynatrace Operator on GKE Autopilot clusters instead.
The Helm repository located in dynatrace/helm-charts is deprecated and will stop receiving updates in a future release! If you are still using it,
please update the URL to dynatrace/dynatrace-operator or switch to the OCI registry-based approach. Update the Helm repository URL with the following commands:
helm repo remove dynatracehelm repo add dynatrace https://raw.githubusercontent.com/Dynatrace/dynatrace-operator/main/config/helm/repos/stable
v1beta3; v1beta4 is now deprecated: The v1beta3 version has been removed from the DynaKube CRD. The v1beta4 version is now deprecated. Migrate your DynaKube resources to v1beta6. For details, see DynaKube API version migration.operator.apparmor, webhook.apparmor, and csidriver.apparmor Helm values are deprecated. On Kubernetes version 1.31+, set AppArmor using appArmorProfile in the podSecurityContext instead. These values will continue to work until Kubernetes version 1.30 reaches end of life (August 2026). See Enable AppArmor for enhanced security for details.dt.kubernetes.cluster.id, dt.kubernetes.workload.kind, and dt.kubernetes.workload.name are deprecated. The new feature.dynatrace.com/enable-attributes-dt.kubernetes feature flag lets you opt out of injecting these by setting it to "false" on the DynaKube. The flag is enabled by default to support a smooth migration to the replacements k8s.cluster.uid, k8s.workload.kind, and k8s.workload.name. In a future release, the flag will default to false, and eventually the deprecated attributes will be removed entirely. Migrate your dashboards, alerts, and queries to the new attributes. For details, see DynaKube feature flags.paasToken field in the dynakube.spec.tokens secret is deprecated. Provide the Operator token in the apiToken field instead.If you upgrade from Dynatrace Operator 1.7.x or earlier, first upgrade to 1.8.x or 1.9.x before moving to 1.10.0. The one-time cleanup for the legacy extensions-controller resources (renamed to extension-controller in 1.8.0) was removed in 1.10.0; skipping the intermediate version leaves stale StatefulSet, Service, and Secret resources in the cluster.
If you deploy Dynatrace Operator on GKE Autopilot using the gke-autopilot.yaml file, switch to Helm deployment before upgrading. This file is no longer published as a release artifact.
Kubernetes recommended labels are now applied consistently across all operator-managed components and other Kubernetes objects. If you select resources by label, verify that your selectors still match after upgrade.
whenUnsatisfiable policy for the ActiveGate TopologySpreadConstraint has changed to ScheduleAnyway. Upgrading to 1.10.0 triggers an ActiveGate restart.