Release 2.13.0 (2026 Q1)

• Kubernetes version: 1.34.3
• The release is supported. See release policy.

Available versions for transition #

¹ When upgrading from 2.11.x to 2.13.x, if logs are collected in the cluster, disable “Local log collection module (Vector)” (shturval-log-collector) before starting the update. Once the update is complete, re-enable this module. Logs collected before the upgrade and placed in the VLogs log storage component will remain available after the upgrade. If metrics are collected in the cluster, disable “Monitoring module control component (VictoriaMetrics Agent)” (shturval-metrics-collector) before starting the update. Once the update is complete, re-enable this module. Metrics collected before the upgrade and stored in the VictoriaMetrics metrics storage component will remain available after the upgrade.

What’s new #

Infrastructure #

• The utility for working with the shtil management cluster has been improved:
  • the ability to install using a YAML manifest from the console is now available;
  • the ability to install via command line flags without using the GUI is now available.
• Added Nvidia GPU integration with automatic node configuration.
• Migrated to Containerd 2.x as container runtime.
• Added checking of Shturval v2 provider hosts for compliance with the minimum system requirements for CPU and Memory.
• Added the ability to manage configuration profiles for client clusters.

GUI #

• The cluster logs GUI has been optimized:
  • logs are grouped by stream;
  • search is available by log parameters with a list of available fields for selection.
• Improved cluster configuration interface:
  • added the ability to set the Ingress DNS record when using the internal Ingress balancer for clusters with providers without Cloud Controller Manager;
  • in client clusters, the default resource configuration has been brought into line with the minimum requirements for cluster installation.
• The ability to use Shturval v2 provider hosts with CPU and Memory below the minimum requirements is blocked.
• Improved Node Configuration Interface (NCI):
  • the ability to configure Kube API Server parameters in the NCI GUI is now available;
  • configuration of all customizable NCI sections is available through modal windows;
  • in the “Container Runtime” section, the ability to specify the address of the Pause image is excluded.
• Optimized management of service operating modes in the cluster via a three-position switch.
• Added display of configuration states of nodes (NodeConfigs) of the cluster.
• Summary reports for all platform clusters have been added to the platform GUI:
  • according to Kyverno security policies;
  • based on the results of scanning container images for vulnerabilities using Trivy.
• Added saving of user settings for filtering, pagination, search queries in the GUI when moving between pages.

Bugs fixed #

• Access error when adding storage in namespace by a user with namespace administrator access rights.
• Applying changes to nodes after updating NCI data in a cluster.
• Incorrect addition of a server to the configuration of an inactive service for time synchronization if both chronyd and ntpd are present on the host.
• Error in accessing changes to recipient data by the cluster administrator in the GUI of the “Alerts” section.
• Incorrect display of values from another service in the available configuration parameters of the GUI service.
• Page loading freezes after editing an alert rule in the GUI.
• Error accessing cluster deletion by cluster administrator.
• Lack of InfraMachineTemplate data in the configuration of a group of Worker nodes of a cluster with the vSphere provider in the GUI.
• Mandatory filling of the shell field when adding a user using NCI in the GUI.
• Incorrect display of NodeConfig status in the node interface when there is no access to cluster node configurations.
• Automatic filling of saved user account data in interface sections with similar fields.

Enterprise ML edition #

• Integrated Kubeflow platform for organizing machine learning (ML) processes in Kubernetes.
• Added MLflow to manage the life cycle of ML models.