Volcengine Kubernetes Engine
Volcengine Kubernetes Engine
Volcengine Kubernetes Engine
Document
Volcengine Kubernetes Engine
What's New
History feature release notes
Kubernetes version release notes
Containerd runtime release notes
Add-on release notes
Monitoring add-ons
Scheduling add-ons
DNS add-ons
Security add-ons
GPU add-ons
Sandbox add-on
Security announcements
Product Announcement
Historical announcements
Overview
Getting started
User guide
Clusters
Creating clusters
Connecting to clusters
Upgrading clusters
Node pool
Creating node pools
Scaling out/in node pools
Nodes
Authorization management
Image
Namespaces
Networking
Cluster network
Container Network
Service
LoadBalancer
CLB
Ingress
APIG Ingress
ALB Ingress
Customize ALB Ingress forwarding rules
Certificate and security
CLB Ingress
Nginx Ingress
Gateway API
Workloads
Creating workloads
Configure workloads
Pods
Configuration management
Image acceleration
Storage
FSX client
EBS Volumes
TOS Volumes
NAS volumes
FSX access methods
NFS access methods
EFS volumes
vePFS volumes
FSX access methods
vePFS Client access methods
NFS access methods
CloudFS volumes
Local volumes
Autoscale
Node autoscaler
Workload autoscaler
Horizontal Pod Autoscaler (HPA)
Cron Horizontal Pod Autoscaler (CronHPA)
Intelligent Horizontal Pod Autoscaler (IHPA)
Horizontal Pod Autoscaler based on event-driven (KEDA)
Add-on management
Network add-ons
Storage add-ons
Monitoring add-ons
Scheduling add-ons
DNS add-ons
GPU add-ons
Core add-ons
Sandbox add-on
Upgrading add-ons
Heterogeneous computing
NVIDIA GPU
mGPU scheduling
RDMA high-performance computing
Scheduling management
Topology-aware scheduling
QoS-aware scheduling
Observability
Log management
Basic observation
Setup & Configuration
Cluster observation
Storage observation
Other observation
Using metrics
Full stack observation
Alert management
Diagnosis and inspection
Node check and self-healing
Application center
Backup center
Stability
Large-scale cluster management
Best practices
Authorization
Clusters
Nodes and node pools
Networking
Storage
Scheduling
Observability
Solution
Security and compliance
API reference
Cluster management
NodePool management
Node management
Virtual nodes management
Add-on management
Tag management
Access policy management
Instance images management
Cloud native observability
Inspection and diagnosis
Task management
Volcengine Container Instance
Appendix
SDK reference
Development guide
Terraform
FAQs
General FAQs
Cluster FAQ
cluster management
cluster network
cluster upgrade
Node and Node pool FAQs
Authorization FAQs
Workload FAQs
Service and Ingress FAQs
Storage FAQs
Heterogeneous computing FAQs
Observability FAQs
Quota FAQs
Add-on FAQs
Cloud-native AI
ServingKit
LLM applications
ComfyUI application
Deploying the Al inference application through Helm
Deepseek practices
Qwen practices
Kimi practice
Seed practices
Best practices
DeepSeek model deployment
GLM model deployment
MiniMax model deployment
Qwen model deployment
Kimi model deployment
Volcengine Container Instance
What's new
Historical feature release notes
Product Announcement
Overview
Billing
User guide
Integrating with VCI
Virtual node
VKE cluster uses virtual nodes
Self-managed/third-party cluster uses virtual node
Creating an instance
General purpose instance
GPU-accelerated computing instance
VCI Agent Sandbox
Startup acceleration
E2B-based management
Upgrade guide
Networking
Monitoring
Scheduling
Container configuration
VCI Image Cache
Best practices
Security and compliance
FAQs
- Documentation
Volcengine Kubernetes EngineVolcengine Kubernetes EngineWhat's NewHistory feature release notesHistory feature release notes (2023)
Copy page
Download PDFHistory feature release notes
History feature release notes (2023)
Copy pageDownload PDFHistory feature release notes (2023)
This document introduces feature-related updates released in 2023 for Volcengine Kubernetes Engine (VKE).
warning
Notes
- Release region refers to the region where the feature was first released. When new release regions are available in the future, the existing region records are not updated. For the regions where features are actually available, please refer to the information on the console pages.
- The historical records of beta testing will not change after beta features are officially released to the public. The iteration information of features is always contained in the latest release notes.
December 2023
Feature name | Description | Region | Release date | Related documentation |
Feature name | Description | Region | Release date | Related documentation |
Full support for Kubernetes version 1.26 | Kubernetes 1.26 in Volcengine Kubernetes Engine is now generally available and officially released following the end of the beta phase. Satisfies users' needs to track native community releases. | China (Beijing) | 2023-12-28 | |
| | China (Guangzhou) | 2023-12-28 | |
| | China (Shanghai) | 2023-12-27 | |
Optimizing node pool scale-out capability | Optimize the scale-out capability of the node pool. By using a balanced scale-out policy, you can avoid using subnets with insufficient remaining IP addresses, thereby reducing the failure rate of node scale-out. | China (Beijing) | 2023-12-28 | None |
| | China (Guangzhou) | 2023-12-28 | |
| | China (Shanghai) | 2023-12-27 | |
Optimizing ResourcePolicy scheduling and supporting the configuration of resource pool ratio thresholds | Beta: When configuring ResourcePolicy scheduling for a workload, you can set a predefined ratio to schedule Pods to Volcengine Container Instance (VCI). ECS computing resources and VCI computing resources can also complement each other, providing better support for workloads with significant traffic peaks and valleys. | China (Beijing) | 2023-12-28 | |
| | China (Guangzhou) | 2023-12-28 | |
| | China (Shanghai) | 2023-12-27 | |
When creating a Service of the LoadBalancer type, you can specify the availability zone of the CLB instance, the instance billing type, and more. | When creating a Service of type LoadBalancer, you can specify the availability zone and billing type of the CLB instance, among other options, thereby supplementing the configuration capabilities available when creating a CLB through VKE and further improving the user experience. | China (Beijing) | 2023-12-28 | |
| | China (Guangzhou) | 2023-12-28 | |
| | China (Shanghai) | 2023-12-27 | |
The console supports batch operation and maintenance on nodes. | The console supports batch operation and maintenance actions for nodes, including batch removal, taking nodes offline, cordoning, setting nodes as schedulable, and more. It also provides one-click navigation to the Elastic Compute Service batch operation and maintenance page. It enhances user experience and meets the need for convenient operation. | China (Beijing) | 2023-12-28 | |
| | China (Guangzhou) | 2023-12-28 | |
| | China (Shanghai) | 2023-12-27 | |
Node pool scaling records can return associated resource IDs and display automatic scaling records triggered by CA and similar mechanisms. | Node pool scaling records can return associated resource IDs. This lets users easily get these IDs from scaling records to complete corresponding initialization configurations. It also displays automatic scaling records triggered by Cluster Autoscaler and similar mechanisms. It addresses user pain points in cloud business migration and enhances user experience. | China (Beijing) | 2023-12-28 | |
| | China (Guangzhou) | 2023-12-28 | |
| | China (Shanghai) | 2023-12-27 | |
The console supports advanced reminders before the expiration of the cluster access configuration using KubeConfig | The console supports advanced expiration reminders for cluster access configurations using KubeConfig, enabling users to take timely action and avoid any impact on business operations. | China (Beijing) | 2023-12-28 | |
| | China (Guangzhou) | 2023-12-28 | |
| | China (Shanghai) | 2023-12-27 | |
Ingress and Service support shared bandwidth packages | Ingress and Service support shared bandwidth packages to improve bandwidth utilization and reduce public network usage costs:
| China (Beijing) | 2023-12-18 | |
| | China (Guangzhou) | 2023-12-18 | |
| | China (Shanghai) | 2023-12-20 | |
Multiple line types for public access for Ingress and Service | [Beta · Apply for Trial] Optimize the configuration of public network traffic entry points for Ingress and Service, supporting multiple line types and configuration of bandwidth limits. Allows users to select Elastic IP Address (EIP) options based on the desired quality and cost in VKE scenarios. | China (Beijing) | 2023-12-18 | |
| | China (Guangzhou) | 2023-12-18 | |
| | China (Shanghai) | 2023-12-20 | |
Enhancing the resource usage display for managed node pools | [Beta] The console now provides enhanced displays of managed node pool resource usage trends, offering an intuitive view to help users easily understand the node pool's overall resource usage. | China (Beijing) | 2023-12-18 | None |
| | China (Guangzhou) | 2023-12-18 | |
| | China (Shanghai) | 2023-12-20 | |
Optimization for the "Degraded" status of add-ons | Optimized the "Degraded" state definition for add-ons. When part of the workload that carries an add-on is abnormal, but the add-on as a whole can still run normally, the add-on can be in the Degraded state. Based on this definition, when an add-on is in the "Degraded" state, its status is adjusted from "Abnormal" to "Normal", allowing upgrade and other O&M operations. | China (Beijing) | 2023-12-18 | None |
| | China (Guangzhou) | 2023-12-18 | |
| | China (Shanghai) | 2023-12-20 | |
November 2023
Feature name | Description | Region | Release date | Related documentation |
Feature name | Description | Region | Release date | Related documentation |
Cluster and container subnets support auxiliary CIDR | Supporting the auxiliary CIDR capability in VPCs ensures network connectivity between Pods in the VKE cluster and the control plane network. This takes effect after you add a VPC subnet created with an auxiliary CIDR to the cluster. It meets the need to expand cluster subnets using auxiliary CIDRs when container workloads face insufficient IP addresses or similar issues. | China (Beijing) | 2023-11-30 | |
| | China (Guangzhou) | 2023-11-29 | |
| | China (Shanghai) | 2023-11-28 | |
Optimizing node pool operations | Optimize node-related operations in the node pool to enhance the user experience. The main optimizations are as follows:
| China (Beijing) | 2023-11-30 | |
| | China (Guangzhou) | 2023-11-29 | |
| | China (Shanghai) | 2023-11-28 | |
"Managed security group" cannot be selected when configuring security groups for a node pool | When configuring a node pool in the console, you cannot select "Managed security group". This prevents users from configuring an unusable managed security group for a VKE node pool, which could otherwise cause node pool scale-out failures and related issues. | China (Beijing) | 2023-11-30 | |
| | China (Guangzhou) | 2023-11-29 | |
| | China (Shanghai) | 2023-11-28 | |
Supports reporting the failure status of key cluster resources to the Cloud Monitor event center | When the status of key resources in the cluster is 'failed', the corresponding event is reported to the Cloud Monitor event center. Through the event-driven mechanism, key events are promptly delivered to users, enabling them to become aware and perform operation and maintenance tasks, thereby improving the observability of the container cluster. | China (Beijing) | 2023-11-30 | |
| | China (Guangzhou) | 2023-11-29 | |
| | China (Shanghai) | 2023-11-28 | |
VPC-CNI cluster supports automatic Pod creation and EIP binding | Clusters using the VPC-CNI network model support automatically creating Elastic IP Addresses (EIP) for Pods and binding them to Pod IPs via annotations. It is suitable for scenarios where individual or a small number of Pods require public network connectivity, and resolves the issue of public network bandwidth contention among Pods. | China (Beijing) | 2023-11-30 | |
| | China (Guangzhou) | 2023-11-29 | |
| | China (Shanghai) | 2023-11-28 | |
Integrate a NAS cache-type file system | Supports using NAS cache type file systems when creating PersistentVolumes statically or dynamically. Meets the requirements for workloads such as machine learning and high-performance computing (HPC) to access data through high-speed file system interfaces. | China (Beijing) | 2023-11-30 | |
| | China (Guangzhou) | 2023-11-29 | |
| | China (Shanghai) | 2023-11-28 | |
ALB Ingress supports configuring redirect rules | When creating an Application Load Balancer (ALB) type Ingress in VKE, you can configure redirect rules to direct requests made to a specific domain, path, or port to another domain, path, or port. The ALB feature has been better adapted to meet user traffic access requirements in scenarios such as domain name changes. | China (Beijing) | 2023-11-15 | |
| | China (Guangzhou) | 2023-11-14 | |
| | China (Shanghai) | 2023-11-13 | |
The node pool supports adding existing elastic scheduled instances. | An Elastic Scheduled Instance (ESI) of Elastic Compute Service (ECS) is a pay-as-you-go instance with delayed delivery. VKE supports filtering and displaying existing ESI instances, and adding them to the node pool. The convenience of using ESI in VKE has been enhanced, meeting users' needs for lower-cost pay-as-you-go resources. The Elastic Scheduled Instance is in the beta phase on the ECS side. If you wish to use it, please submit an application via ECS. | China (Beijing) | 2023-11-15 | None |
| | China (Guangzhou) | 2023-11-14 | |
| | China (Shanghai) | 2023-11-13 | |
Node now supports the c3i.48xlarge specification | Node now supports the c3i.48xlarge specification in the ECS compute-optimized c3i instance family. Meets user requirements for compute-optimized instances of medium and large sizes. | China (Beijing) | 2023-11-15 | None |
| | China (Guangzhou) | 2023-11-14 | |
| | China (Shanghai) | 2023-11-13 | |
Support self-healing capability for GPU nodes | [Beta · Apply for Trial]The AI suite enables users to configure self-healing rules for GPU node pools. It automatically performs fault detection and notification, and achieves self-healing through rules such as stopping scheduling, evicting Pods, restarting ECS, and other rules. It helps users improve the efficiency of resolving GPU issues and speeds up the recovery of AI services. | China (Beijing) | 2023-11-15 | |
| | China (Guangzhou) | 2023-11-14 | |
| | China (Shanghai) | 2023-11-13 | |
The scheduler-plugin add-on supports feature-level switches and parameter configuration | The scheduler-plugin add-on allows users to enable or disable scheduling policies during and after installation, and supports customizing the weights of certain scheduling policies. It meets user requirements for flexible management and configuration of advanced scheduling policies. | China (Beijing) | 2023-11-15 | |
| | China (Guangzhou) | 2023-11-14 | |
| | China (Shanghai) | 2023-11-13 | |
Batch Computing Suite console capabilities upgrade | [Beta] The batch computing suite console now supports configuring VCI CPU and VCI GPU resource specifications for queues. The overview page and queue list also add data displays, including queue job time statistics and more. It provides batch computing users with an improved console experience, enabling more convenient use of VCI elastic container resources. | China (Beijing) | 2023-11-15 | None |
| | China (Guangzhou) | 2023-11-14 | |
| | China (Shanghai) | 2023-11-13 | |
Prometheus monitoring dashboard optimization | Optimized the clustering and navigation of Prometheus monitoring dashboards in VKE.Added dedicated dashboards for CoreDNS, NodeLocalDNS, and Namespace monitoring. Made adjustments to other existing dashboards. This gives VKE users a monitoring experience better aligned with O&M needs. It also fills key observation blind spots in the data plane. | China (Beijing) | 2023-11-15 | None |
| | China (Guangzhou) | 2023-11-14 | |
| | China (Shanghai) | 2023-11-13 | |
New standalone virtual node module in the console | [Beta · Apply for Trial] The VKE console has added a virtual node module directory, supporting viewing, creating, and deleting operations on virtual nodes. For self-managed Kubernetes clusters, registering managed virtual nodes in the cloud offers a convenient way to use Volcengine Container Instance and provides an improved console visualization experience. | China (Beijing) | 2023-11-15 | None |
| | China (Guangzhou) | 2023-11-14 | |
| | China (Shanghai) | 2023-11-13 | |
October 2023
Feature name | Description | Region | Release date | Related documentation |
Feature name | Description | Region | Release date | Related documentation |
Support granting cluster RBAC permissions to IAM roles | Enhanced support for granting cluster RBAC permissions to IAM roles. This enables users to log in via role-based SSO from an identity provider (IdP) and use Volcengine Kubernetes Engine (VKE) capabilities to support user scenarios. | China (Beijing) | 2023-11-01 | |
| | China (Guangzhou) | 2023-10-30 | |
| | China (Shanghai) | 2023-10-27 | |
Support custom metadata name prefixes | Volcengine Kubernetes Engine configures a prefix for the metadata name (metadata.name) of nodes in a node pool. Users can use the node's metadata.name to identify resources in external systems such as monitoring and alert systems, and to quickly locate the corresponding node resources. | China (Beijing) | 2023-11-01 | |
| | China (Guangzhou) | 2023-10-30 | |
| | China (Shanghai) | 2023-10-27 | |
NetworkPolicy capability support for Kubernetes 1.20 VPC-CNI clusters | [Beta·Apply for Trial]Supports the use of the NetworkPolicy feature in Kubernetes 1.20 VPC-CNI clusters. Provide policy-based network control for Kubernetes 1.20 VPC-CNI clusters to enhance the security capabilities of Pod networking within the cluster. | China (Beijing) | 2023-11-01 | None |
| | China (Guangzhou) | 2023-10-30 | |
| | China (Shanghai) | 2023-10-27 | |
Metric collection for add-ons now supports CoreDNS and NodeLocalDNS | Add-on metric collection now supports the core-dns and node-local-dns add-ons, allowing users to configure DNS metric collection as needed and use the collected metrics to build DNS monitoring dashboards and alert rules, thereby enhancing add-on operation and maintenance capabilities. | China (Beijing) | 2023-11-01 | |
| | China (Guangzhou) | 2023-10-30 | |
| | China (Shanghai) | 2023-10-27 | |
Cloud Native AI Suite beta now open | [beta] The Cloud-Native AI Suite is a VKE-based suite that supports unified management of infrastructure requirements for artificial intelligence (AI) platforms. It reduces learning and add-on management costs for users in AI scenarios using VKE. | China (Beijing) | 2023-11-01 | |
| | China (Guangzhou) | 2023-10-30 | |
| | China (Shanghai) | 2023-10-27 | |
CA adds support for GPU scale-in threshold configuration | Cluster Autoscaler (CA) now supports GPU scale-in threshold configuration, enabling timely scale-in of GPU resources when they are no longer in use and improving GPU resource utilization efficiency. | China (Beijing) | 2023-11-01 | |
| | China (Guangzhou) | 2023-10-30 | |
| | China (Shanghai) | 2023-10-27 | |
Supports configuring independent health check parameters for different backend server groups in ALB Ingress | Supports configuring independent health check parameters for different backend server groups of ALB type Ingress (routing rules) using annotations. Meets differentiated configuration requirements for Ingress. | China (Beijing) | 2023-10-17 | |
| | China (Guangzhou) | 2023-10-17 | |
| | China (Shanghai) | 2023-10-13 | |
After the cluster connects to VMP, it supports selectively enabling reporting of monitoring metrics for add-ons and allows customization and configuration of the metric collection scope for add-ons. | After a cluster connects to Volcengine Managed Service for Prometheus (VMP), you can selectively enable monitoring metric reporting for add-ons and customize the metric collection scope for add-ons. The system provides recommended best practices for cluster add-on monitoring metrics. These features help users reduce the cost of using Volcengine Managed Service for Prometheus and enable more effective observation of add-on metrics. | China (Beijing) | 2023-10-17 | |
| | China (Guangzhou) | 2023-10-17 | |
| | China (Shanghai) | 2023-10-13 | |
Supports configuring JSON log collection rules using CRD. | This feature supports configuring JSON log collection rules via CRD, enabling users to ingest more log types and obtain more complete data during alert troubleshooting. | China (Beijing) | 2023-10-17 | |
| | China (Guangzhou) | 2023-10-17 | |
| | China (Shanghai) | 2023-10-13 | |
Console Pod port adaptation to string data type | When configuring ports for workloads, Services, and other components in the console, both int type (port numbers) and String type (port names) are supported, in alignment with native Kubernetes capabilities. | China (Beijing) | 2023-10-17 | |
| | China (Guangzhou) | 2023-10-17 | |
| | China (Shanghai) | 2023-10-13 | |
Adding a managed node pool with fault detection capability for GPU nodes | [Beta] The managed node pool feature has been added. The node pool helps users enhance fault management capability for GPU-based compute resources simply and efficiently. It also improves node pool resource utilization and simplifies daily operation and maintenance tasks. | China (Beijing) | 2023-10-17 | None |
| | China (Guangzhou) | 2023-10-17 | |
| | China (Shanghai) | 2023-10-13 | |
Optionally disable auto-refresh on the console workload list page | On the console workload list page, you can optionally disable auto-refresh to reduce the pressure on the API Server and other control plane add-ons caused by frequent frontend requests. Provide underlying capability support for obtaining workloads in all namespaces. | China (Beijing) | 2023-10-17 | None |
| | China (Guangzhou) | 2023-10-17 | |
| | China (Shanghai) | 2023-10-13 | |
Retain configuration for some user-defined parameters of core-dns and other add-ons | Some configuration parameters of add-ons such as core-dns support retaining user-defined configurations during configuration changes, add-on upgrades, and similar scenarios. It has helped users reduce the complexity of add-on operation and maintenance. | China (Beijing) | 2023-10-17 | None |
| | China (Guangzhou) | 2023-10-17 | |
| | China (Shanghai) | 2023-10-13 | |
September 2023
Feature name | Description | Region | Release date | Related documentation |
Feature name | Description | Region | Release date | Related documentation |
VPC-CNI network add-on supports binding Elastic IP Addresses | [Beta · Apply for Trial] The VPC-CNI network add-on supports binding an Elastic IP Address. Provide independent public network ingress and egress for Pods to support scenarios where a single Pod or a small number of Pods need to communicate over the public network, and to resolve bandwidth contention issues among Pods. | China (Beijing) | 2023-09-26 | None |
| | China (Guangzhou) | 2023-09-26 | |
| | China (Shanghai) | 2023-09-25 | |
Cluster supports modifying the API Server subnet | [Beta · Apply for Trial] After creating a cluster, you can add API Server subnets. By adding API Server subnet capabilities, you can still add API Server subnets to existing single-subnet clusters, enabling disaster recovery across availability zones. | China (Beijing) | 2023-09-26 | None |
| | China (Guangzhou) | 2023-09-26 | |
| | China (Shanghai) | 2023-09-25 | |
The cluster supports upgrading the VKE patch version within a specified Kubernetes version number | [Beta · Apply for Trial] When upgrading a cluster's Kubernetes version, you can first specify the latest VKE patch version (in the format vke.(n), and then upgrade the Kubernetes minor version as needed. This helps control cluster upgrade risk. | China (Beijing) | 2023-09-26 | |
| | China (Guangzhou) | 2023-09-26 | |
| | China (Shanghai) | 2023-09-25 | |
Allow ignoring checks with warning results to perform a Kubernetes version upgrade | [Beta · Apply for Trial]Previously, when upgrading the Kubernetes version of a cluster or node, the upgrade would be interrupted by default if any alert-level check items were detected. This has been changed to allow you to ignore the check results and continue upgrading the Kubernetes version. To avoid user upgrades being blocked by overly strict monitoring rules, users should fully assess the risks before proceeding with the upgrade as usual. | China (Beijing) | 2023-09-26 | |
| | China (Guangzhou) | 2023-09-26 | |
| | China (Shanghai) | 2023-09-25 | |
Supporting self-healing checks for GPU nodes | [Beta·Apply for Trial] Supports self-healing for common issue checks on GPU nodes. By leveraging fault self-healing capabilities, you can reduce the time required to detect GPU failures and minimize their impact on training jobs, while improving the efficiency of resolving GPU issues. | China (Beijing) | 2023-09-26 | |
| | China (Guangzhou) | 2023-09-26 | |
| | China (Shanghai) | 2023-09-25 | |
The MVP version of the cloud-native batch computing suite is now open for beta testing | [Beta] The cloud-native batch computing suite provides integrated batch computing product capabilities based on VKE. Leveraging Elastic Compute Service and Volcengine Container Instance managed by VKE, it offers users simple and easy-to-use batch computing capabilities, reducing complexity in job scheduling and resource management. | China (Beijing) | 2023-09-12 | |
| | China (Guangzhou) | 2023-09-11 | |
| | China (Shanghai) | 2023-09-05 | |
Allow configuring ACL via Annotation when creating or updating a CLB type Service | When creating or updating a CLB type Service, you can configure the access control policy group (ACL) using annotations. The ability for users to manage CLB-type services through VKE has been optimized, making it easier to control access permissions. | China (Beijing) | 2023-09-12 | |
| | China (Guangzhou) | 2023-09-11 | |
| | China (Shanghai) | 2023-09-05 | |
Service annotation supports enabling Proxy Protocol source IP passthrough | Service supports enabling Proxy Protocol source IP passthrough by configuring annotations. Provides the capability to obtain source IP addresses to meet user requirements for data analytics, security auditing of requests, security attack tracing, and other business scenarios. | China (Beijing) | 2023-09-12 | |
| | China (Guangzhou) | 2023-09-11 | |
| | China (Shanghai) | 2023-09-05 | |
Configuring environment variables to collect VKE container logs in Torch Log Service | Beta: Supports configuring log collection rules by setting environment variables in the container. Configurable items include log type, log file path, Torch Log Service project and topic, maximum retention period, number of shards, and more. This feature meets diverse user needs for log collection and lets you configure log collection rules more quickly and conveniently. | China (Beijing) | 2023-09-12 | |
| | China (Guangzhou) | 2023-09-11 | |
| | China (Shanghai) | 2023-09-05 | |
August 2023
Feature name | Description | Region | Release date | Related documentation |
Feature name | Description | Region | Release date | Related documentation |
The rdma-device-plugin add-on supports collection of RDMA network interface card metrics within containers. | [Beta · Apply for Trial]After users install the rdma-device-plugin add-on, a series of observability capabilities are provided, including RDMA metric collection, storage, alert configuration, and more. Enhance the user experience with RDMA devices, assist users in tuning and troubleshooting RDMA devices, and enable them to better use RDMA devices to improve application performance. | China (Beijing) | 2023-08-29 | |
| | China (Guangzhou) | 2023-08-23 | |
| | China (Shanghai) | 2023-08-21 | |
nvidia-device-plugin add-on outputs GPU Xid logs | Supports outputting GPU Xid-related information to the container log file/var/log/xid.log. GPU Xid logs enhance the user experience when using GPU devices. By monitoring the occurrence time and specific details of Xid events through GPU Xid logs, users can troubleshoot GPU device failures and improve stability when running training jobs on GPU devices. | China (Beijing) | 2023-08-29 | None |
| | China (Guangzhou) | 2023-08-23 | |
| | China (Shanghai) | 2023-08-21 | |
Provide a monitoring dashboard for training jobs | [Beta · Apply for Trial] Provides users with a dedicated monitoring dashboard for AI training Jobs, supporting observation of metrics such as GPU usage, GPU anomalies, GPU status, Job status, and work quantity from different perspectives, including cluster and node. Enhance users' ability to monitor GPUs in AI training scenarios to achieve better fault alerting, cost optimization, resource management, and training job management. | China (Beijing) | 2023-08-29 | |
| | China (Guangzhou) | 2023-08-23 | |
| | China (Shanghai) | 2023-08-21 | |
Supports one-click enabling of VMP alerts | Provides a feature to enable Prometheus alerts for the cluster with one click. Optimize the experience for VKE users using Volcengine Managed Service for Prometheus. Make it easier to enable and manage VMP alert capabilities. | China (Beijing) | 2023-08-29 | None |
| | China (Guangzhou) | 2023-08-23 | |
| | China (Shanghai) | 2023-08-21 | |
The VPC-CNI network model cluster supports using Trunk ENI as the primary network interface for Pods | [Beta · Apply for Trial] Supports enabling Trunk ENI mode when creating a cluster using the VPC-CNI network model. Once enabled, you can bind Branch ENIs to Pods within Trunk ENI nodes, providing each Pod with independent network interface resources and allowing Pods to specify subnets and Security Groups. Address network isolation requirements between Pods and more conveniently integrate Pods with other Volcano Engine services, such as LoadBalancer. | China (Beijing) | 2023-08-29 | |
| | China (Guangzhou) | 2023-08-23 | |
| | China (Shanghai) | 2023-08-21 | |
Support for Kubernetes version 1.26 | Beta: Clusters support Kubernetes 1.26, enabling users to keep up with releases from the upstream community. | China (Beijing) | 2023-08-29 | |
| | China (Guangzhou) | 2023-08-23 | |
| | China (Shanghai) | 2023-08-21 | |
Support NUMA-aware scheduling | [Beta · Apply for Trial] You can enable NUMA-aware scheduling for node pools. After enabling this feature, the scheduler will be aware of the NUMA micro-topology of the node, allowing containers to be scheduled and bound to CPUs on a single NUMA node at the container level. Improve container runtime performance and enhance the experience of latency-sensitive workloads. | China (Beijing) | 2023-08-29 | |
| | China (Guangzhou) | 2023-08-23 | |
| | China (Shanghai) | 2023-08-21 | |
Expose the S0 hash information of RDMA network cards on HPC nodes | [Beta · Apply for Trial] On VKE high-performance computing (HPC) nodes, you can use the node label (topology.vke.volcengine.com/switch) to automatically expose the merged hash values of multiple S0 switches connected to the node's RDMA network card. This helps users determine whether multiple nodes are connected to the same S0 switch group by using the S0 hash value. When scheduling Pods, this enables the selection of nodes with a shorter RDMA topology distance, thereby improving GPU communication performance between Pods. | China (Beijing) | 2023-08-29 | None |
| | China (Guangzhou) | 2023-08-23 | |
| | China (Shanghai) | 2023-08-21 | |
Support dispersing load hotspots through rescheduling | [Beta · Apply for Trial] Provides the capability to disperse and reschedule load hotspots, detect changes in node load within the cluster, and reschedule Pods on nodes that exceed the safe load threshold. | China (Beijing) | 2023-08-29 | None |
| | China (Guangzhou) | 2023-08-23 | |
| | China (Shanghai) | 2023-08-21 | |
Providing cluster fault diagnosis tools | [Beta · Apply for Trial] Provides a cluster fault diagnosis tool that performs fault diagnosis on node pools, nodes, or pods, quickly locates the cause of faults, reduces the time required for fault diagnosis and recovery, and improves overall business continuity. | China (Beijing) | 2023-08-10 | |
| | China (Guangzhou) | 2023-08-09 | |
| | China (Shanghai) | 2023-08-08 | |
Provide cluster inspection tools and support daily cluster inspection | [Beta·Apply for Trial] Add cluster inspection feature, supporting custom inspection rules and performing periodic inspections of clusters, nodes, external dependencies, key components, quotas (Quota), and more. Identify potential risks in clusters early, reduce the failure rate, and improve overall cluster operational efficiency. | China (Beijing) | 2023-08-10 | |
| | China (Guangzhou) | 2023-08-09 | |
| | China (Shanghai) | 2023-08-08 | |
Add-on management now supports the rdma-device-plugin add-on | Beta: Added the rdma-device-plugin add-on. Supports modular installation of RDMA device add-ons. Enables convenient use of RDMA capabilities in both exclusive and shared modes. This improves user experience. | China (Beijing) | 2023-08-10 | |
| | China (Guangzhou) | 2023-08-09 | |
| | China (Shanghai) | 2023-08-08 | |
Workloads support writing ZoneId to Pod Annotation | After the Pod is scheduled to a node, the availability zone (AZ) information for that node is propagated to the Pod and written into the Pod annotation. This enables business operations to optimize service performance based on AZ information. | China (Beijing) | 2023-08-10 | None |
| | China (Guangzhou) | 2023-08-09 | |
| | China (Shanghai) | 2023-08-08 | |
Node has added support for the g3a and c3a specifications | [Beta · Apply for Trial] Node now supports Elastic Compute Service (ECS) general-purpose g3a and compute-optimized c3a specifications. Suitable for scenarios that require high network and storage I/O performance. These two specifications are beta specifications on both the ECS and VKE platforms. If you need to use them, please submit applications for both ECS and VKE simultaneously. | China (Beijing) | 2023-08-10 | None |
The maximum length of node deployment scripts has been expanded to 8 KB | [Beta · Apply for Trial] The length limit for custom script commands deployed on the node has been increased from 1 KB to 8 KB to meet user experience requirements. | China (Beijing) | 2023-08-10 | None |
| | China (Guangzhou) | 2023-08-09 | |
| | China (Shanghai) | 2023-08-08 | |
Prometheus monitoring adds GPU power and temperature metrics | Prometheus monitoring adds GPU power and temperature metrics, and supports filtering GPU instances. Complete GPU observation scenarios and metrics to enhance the role of the pod GPU monitoring dashboard in the troubleshooting process. | China (Beijing) | 2023-08-10 | |
| | China (Guangzhou) | 2023-08-09 | |
| | China (Shanghai) | 2023-08-08 | |
July 2023
Feature name | Description | Region | Release date | Related documentation |
Feature name | Description | Region | Release date | Related documentation |
Adding specific labels to ECS nodes used in VKE | On Elastic Compute Service (ECS) instances, add the volc:vke:nodepool-id label to ECS nodes used in Volcengine Kubernetes Engine (VKE) clusters. This enables users to quickly identify which ECS instances are VKE nodes and provides better support for scenarios at the node pool level, such as resource management, cost analysis, and more. | China (Beijing) | 2023-07-20 | None |
| | China (Guangzhou) | 2023-07-19 | |
| | China (Shanghai) | 2023-07-18 | |
Nodes of the preemptible instance type report corresponding events before being preempted. | [Beta · Apply for Trial]ECS preemptible instances, due to their inherent characteristics, may be forcibly reclaimed due to insufficient inventory. In VKE, this type of node reports corresponding events five minutes before being reclaimed and again when being reclaimed. This enables users to promptly become aware of the release of preemptible instance type nodes and take appropriate measures. | China (Beijing) | 2023-07-20 | None |
| | China (Guangzhou) | 2023-07-19 | |
| | China (Shanghai) | 2023-07-18 | |
An Elastic IP Address automatically bound to a node is released simultaneously with the node | [Beta · Apply for Trial] After the node pool enables the 'automatic node binding to Elastic IP Address' feature, deleting or scaling in nodes will simultaneously release the Elastic IP Address bound to those nodes. Meets the requirements in autoscaler scenarios to bind EIPs to nodes according to node pool configuration during scale-out, and to release EIPs synchronously during node scale-in, thereby improving user experience. | China (Beijing) | 2023-07-20 | |
| | China (Guangzhou) | 2023-07-19 | |
| | China (Shanghai) | 2023-07-18 | |
When the resource inventory in the node pool's availability zone is insufficient, resources will be automatically scheduled to a new availability zone. | When creating or scaling out a node pool, if the ECS resources of the node pool are insufficient in an availability zone, the system will automatically provision ECS resources from other availability zones within the subnet where the node pool is located. This prevents failures in creating or scaling out node pools caused by resource issues in a single availability zone. | China (Beijing) | 2023-07-20 | None |
| | China (Guangzhou) | 2023-07-19 | |
| | China (Shanghai) | 2023-07-18 | |
Display associated resource information after the cluster has been created | After the cluster is created, you can view the IDs of associated resources including Security Groups, NAT Gateways, and LoadBalancer (CLB). You can also quickly navigate to their respective cloud resource details pages. | China (Beijing) | 2023-07-20 | |
| | China (Guangzhou) | 2023-07-19 | |
| | China (Shanghai) | 2023-07-18 | |
Provide a dashboard for monitoring Image Lazy Loading metrics | [Beta · Apply for Trial] Provides a dashboard for Nydus Image Lazy Loading capability metrics, improving the observability of the Nydus capability. | China (Beijing) | 2023-07-20 | |
| | China (Guangzhou) | 2023-07-19 | |
| | China (Shanghai) | 2023-07-18 | |
Nodes now also support ECS specifications | Node now additionally supports the Big Data d2c instance type of Elastic Compute Service (ECS). Supports scenarios such as big data computing (for example, Hadoop), machine learning, and log processing. This specification is a beta specification on the ECS side. If you need to use it, please submit an application on the ECS side. | China (Beijing) | 2023-07-11 | None |
| | China (Guangzhou) | 2023-07-10 | |
| | China (Shanghai) | 2023-07-07 | |
Support for NetworkPolicy | [Beta · Apply for Trial] VPC-CNI clusters support NetworkPolicy. NetworkPolicy provides policy-based network control, allowing you to manage Pod-level network traffic based on IP address or port, which further ensures network security. | China (Beijing) | 2023-07-11 | |
| | China (Guangzhou) | 2023-07-10 | |
| | China (Shanghai) | 2023-07-07 | |
HPA supports the network throughput metric | [Beta · Apply for Trial] Supports workload metric autoscaler (HPA) based on network throughput metrics. Meets the need for bandwidth-sensitive workloads to perform horizontal Pod autoscaling based on network throughput metrics, thereby enhancing business elasticity. | China (Beijing) | 2023-07-11 | |
| | China (Guangzhou) | 2023-07-10 | |
| | China (Shanghai) | 2023-07-07 | |
The prometheus-agent add-on supports ECS/VCI proxy authentication. | The prometheus-agent add-on uses ECS/VCI proxy authentication, allowing users to install prometheus-agent during cluster creation or add-on management without entering authentication information. | China (Beijing) | 2023-07-11 | |
| | China (Guangzhou) | 2023-07-10 | |
| | China (Shanghai) | 2023-07-07 | |
Supporting Prometheus workspace configuration | You can configure a Prometheus workspace at the cluster level. This allows all observational data from the same cluster to be aggregated into a single Prometheus workspace, meeting the query requirements for various observation scenarios. | China (Beijing) | 2023-07-11 | |
| | China (Guangzhou) | 2023-07-10 | |
| | China (Shanghai) | 2023-07-07 | |
Namespaces support custom descriptions | Namespace supports adding descriptions and search capabilities based on them, enhancing the namespace management experience. | China (Beijing) | 2023-07-11 | |
| | China (Guangzhou) | 2023-07-10 | |
| | China (Shanghai) | 2023-07-07 | |
Upgrading Containerd to version v1.6.21 | Upgrade the Containerd version of the cluster to v1.6.21 to enhance stability. | China (Beijing) | 2023-07-11 | |
| | China (Guangzhou) | 2023-07-10 | |
| | China (Shanghai) | 2023-07-07 | |
Increase operation restrictions for NAT gateways associated with VKE | When NAT gateways are created in association with Volcengine Kubernetes Engine and used in a cluster, users are prohibited from manually deleting them.Accidental deletion of NAT gateway instances may cause abnormal business operations. | China (Beijing) | 2023-07-11 | |
| | China (Guangzhou) | 2023-07-10 | |
| | China (Shanghai) | 2023-07-07 | |
Supporting multiple Ingress controllers | You can deploy multiple Nginx Ingress Controllers in the cluster and use different controllers to separate internal and external network traffic or to distinguish ingress traffic for different services. | China (Beijing) | 2023-07-11 | |
| | China (Guangzhou) | 2023-07-10 | |
| | China (Shanghai) | 2023-07-07 | |
June 2023
Feature name | Description | Region | Release date | Related documentation |
Feature name | Description | Region | Release date | Related documentation |
Supporting CloudFS CSI add-on | [Beta · Apply for Trial] Added csi-cloudfs add-on for using CloudFS via CSI. CloudFS supports standard HDFS protocol access and transparent access mode for data lakes, providing low-cost, high-performance, high-throughput, and highly available access to big data files. | China (Beijing) | 2023-06-26 | |
| | China (Guangzhou) | | |
| | China (Shanghai) | 2023-06-21 | |
Kubernetes upgrade check supports incompatible API detection | [Beta · Apply for Trial]Kubernetes upgrade checks support incompatible API detection, preventing workload incompatibility caused by version upgrades, providing more comprehensive upgrade checks, and improving upgrade success rates. | China (Beijing) | 2023-06-26 | |
| | China (Guangzhou) | | |
| | China (Shanghai) | 2023-06-21 | |
Node pool supports cloning capability | Node pools support creating new custom node pools by cloning, which improves user convenience in node pool creation and console operation efficiency. | China (Beijing) | 2023-06-26 | |
| | China (Guangzhou) | | |
| | China (Shanghai) | 2023-06-21 | |
Optimize the server list and filtering feature when adding existing nodes | When adding existing nodes, searching and filtering by Elastic Compute Service name or ID is supported to enhance the user experience. | China (Beijing) | 2023-06-26 | None |
| | China (Guangzhou) | | |
| | China (Shanghai) | 2023-06-21 | |
Updating the installation recommendations for the vci-virtual-kubelet add-on | The vci-virtual-kubelet add-on has changed from system-required installation to recommended installation. For users who do not use Volcengine Container Instance (VCI), resource consumption can be reduced, and issues caused by Daemonset scheduling to virtual nodes can be avoided. | China (Beijing) | 2023-06-26 | |
| | China (Guangzhou) | | |
| | China (Shanghai) | 2023-06-21 | |
Full support for Kubernetes 1.24 | The public beta for Kubernetes 1.24 on Volcengine Kubernetes Engine has ended, and the version has now been officially released. This meets users' needs to keep up with native community releases. | China (Beijing) | 2023-06-25 | |
| | China (Guangzhou) | | |
| | China (Shanghai) | | |
Nodes now support ECS specifications | Nodes now support Elastic Compute Service (ECS) general-purpose g3i, compute-optimized c3i, and memory-optimized r3i specifications. Meet the usage requirements of different users for various ECS specifications in different scenarios. The above specifications are in beta on ECS. If you need to use them, please submit an application through ECS. | China (Beijing) | 2023-06-13 | None |
Automatic installation and startup of the Fabric Manager Service when high-performance worker nodes are added to VKE | Pre-install NVIDIA Fabric Manager startup scripts to enable multi-GPU interconnection via NVSwitch among A100 and similar models. Resolve scenarios in which instances do not function as expected and optimize user experience. | China (Beijing) | 2023-06-13 | |
| | China (Guangzhou) | 2023-06-12 | |
| | China (Shanghai) | 2023-06-06 | |
May 2023
Feature name | Description | Region | Release date | Related documentation |
Feature name | Description | Region | Release date | Related documentation |
Clusters support custom local domain names | When creating a cluster, the system supports modifying the default local domain name of the DNS server within the cluster. This ensures that Service names are globally unique within the cluster and supports scenarios involving DNS domain name resolution across multiple clusters. | China (Beijing) | 2023-05-29 | |
| | China (Guangzhou) | 2023-05-26 | |
| | China (Shanghai) | 2023-05-25 | |
Creating a node pool that supports local disks | [Beta · Apply for Trial] Local disk mounting is supported when creating a node pool. Local disks provide nodes with access to local storage, offering advantages such as low latency, high random IOPS, high throughput, and high cost-effectiveness. | China (Beijing) | 2023-05-29 | |
| | China (Guangzhou) | 2023-05-26 | |
| | China (Shanghai) | 2023-05-25 | |
Add a notice in the console indicating that VMP is now fully available for commercial use | Volcengine Managed Service for Prometheus (VMP) will begin charging on May 29, 2023. Some features of Volcengine Kubernetes Engine use VMP capabilities. If you do not need to use this part of the feature and have installed the prometheus-agent add-on, please uninstall the add-on to avoid incurring charges. | China (Beijing) | 2023-05-29 | |
| | China (Guangzhou) | 2023-05-26 | |
| | China (Shanghai) | 2023-05-25 | |
Ingress and Service support multi-availability zone capability for ALB and CLB | Some Ingresses and Services support multiple availability zones, enhancing the high availability of containerized business applications.
| China (Beijing) | 2023-05-11 | |
| | China (Guangzhou) | 2023-05-10 | |
| | China (Shanghai) | 2023-05-09 | |
Connecting to projects | [Beta · Apply for Trial] Cluster resources are now connected to projects (Project), allowing users to manage clusters across all projects, and also allowing users to specify the project for the ECS instance corresponding to each node in the node pool configuration. This meets users' requirements for fine-grained permission management in Volcengine Kubernetes Engine by narrowing the permission scope to the project level. | China (Beijing) | 2023-05-11 | |
| | China (Guangzhou) | 2023-05-10 | |
| | China (Shanghai) | 2023-05-09 | |
Pod networking supports the VPC-CNI dual network interface model | [Beta · Apply for Trial] Pod networking is implemented based on Trunk ENI technology, enabling the VPC-CNI dual NIC model. Based on the dual network interface card model, different network planes are planned for business purposes. It supports binding auxiliary network interfaces of Pods to security groups, providing NIC-level network traffic isolation for Pods within the cluster. | China (Beijing) | 2023-05-11 | None |
| | China (Guangzhou) | 2023-05-10 | |
| | China (Shanghai) | 2023-05-09 | |
Prometheus monitoring module is now fully open | The Prometheus monitoring feature and the prometheus-agent add-on are now fully available, as the beta phase has ended. | China (Beijing) | 2023-05-11 | |
| | China (Guangzhou) | 2023-05-10 | |
| | China (Shanghai) | 2023-05-09 | |
Prometheus Agent supports a high availability configuration | The prometheus-agent add-on supports high availability configuration to meet the needs of large-scale cluster monitoring and collection scenarios. | China (Beijing) | 2023-05-11 | |
| | China (Guangzhou) | 2023-05-10 | |
| | China (Shanghai) | 2023-05-09 | |
P2P acceleration feature public beta | [Public Beta] The P2P-accelerator add-on has ended its invitation-only beta and transitioned to Public Beta, now providing P2P image acceleration capability. The P2P acceleration feature can help users significantly improve the speed and availability when pulling container images with high concurrency. | China (Beijing) | 2023-05-11 | |
| | China (Guangzhou) | 2023-05-10 | |
| | China (Shanghai) | 2023-05-09 | |
CentOS-compatible version of the veLinux operating system is now fully available | Support for the veLinux CentOS compatible operating system on node images of certain specifications is no longer in beta and is now fully available. | China (Beijing) | 2023-05-11 | |
| | China (Guangzhou) | 2023-05-10 | |
| | China (Shanghai) | 2023-05-09 | |
Custom node pools support adding existing nodes | [Beta] Custom node pool supports adding existing nodes under certain conditions. It meets the requirements of business scenarios such as resource reuse, node preheating, and others. | China (Beijing) | 2023-05-11 | |
| | China (Guangzhou) | 2023-05-10 | |
| | China (Shanghai) | 2023-05-09 | |
Nodes support the policy for pulling images in parallel | Nodes enable the parallel image pull policy by default. When there are multiple business Pods on a node, the image pull time for Pods is reduced, which mitigates the issue of failures to add existing nodes caused by image pull timeouts. | China (Beijing) | 2023-05-11 | None |
| | China (Guangzhou) | 2023-05-10 | |
| | China (Shanghai) | 2023-05-09 | |
April 2023
Feature name | Description | Region | Release date | Related documentation |
Feature name | Description | Region | Release date | Related documentation |
Add-on support for NodeLocal DNS capability | NodeLocal DNSCache runs as a DaemonSet on cluster nodes, operating a DNS cache proxy to improve DNS performance in the cluster. Containers can access the NodeLocalCache add-on on the local node to perform DNS resolution, thereby improving cluster DNS domain name resolution performance. | China (Beijing) | 2023-04-24 | |
| | China (Guangzhou) | 2023-04-23 | |
| | China (Shanghai) | 2023-04-21 | |
ALB Ingress supports configuring rate limiting for forwarding rules | [Beta · Apply for Trial] ALB Ingress supports configuring different rate limiting rules for different Services based on priority, helping prevent overall Service unavailability caused by excessive requests to specific business functions, and meeting user requirements for granular traffic control and traffic grouping restrictions. | China (Beijing) | 2023-04-24 | |
| | China (Guangzhou) | 2023-04-23 | |
| | China (Shanghai) | 2023-04-21 | |
When creating a VPC-CNI cluster with zero nodes, some add-ons are deployed to VCI by default. | When creating a cluster using the VPC-CNI network model, if nodes are not created synchronously or the number of nodes is zero, some add-ons are deployed by default as Volcengine Container Instances (VCI) to optimize the cluster creation and usage experience in Serverless Kubernetes scenarios. | China (Beijing) | 2023-04-24 | |
| | China (Guangzhou) | 2023-04-23 | |
| | China (Shanghai) | 2023-04-21 | |
Adding a risk warning when removing a node | When removing a node, explicitly warn users about the potential risks, including reinstalling the node's operating system, erasing data on the system volume, and other possible risks. | China (Beijing) | 2023-04-24 | |
| | China (Guangzhou) | 2023-04-23 | |
| | China (Shanghai) | 2023-04-21 | |
Support for label-based bill splitting | Volcengine Kubernetes Engine (managed version) supports label-based bill splitting. With the enhanced bill splitting feature, users can view split bills in the Volcano Engine console. | China (Beijing) | 2023-04-24 | |
| | China (Guangzhou) | 2023-04-23 | |
| | China (Shanghai) | 2023-04-21 | |
mGPU supports multi-card sharing | [Beta] In mGPU scenarios, a single container can request multiple fractional GPU resources (computing power and/or memory). Support for multi-card sharing of computing power and memory has been implemented, improving the utilization of fragmented GPU resources. | China (Beijing) | 2023-04-11 | None |
| | China (Guangzhou) | 2023-04-10 | |
| | China (Shanghai) | 2023-04-07 | |
Optimizing custom image tools | [Beta · Apply for Trial] Optimize the custom image tool and provide automatic preheating capability. The custom image tool is a one-stop tool that enables you to complete image building, scanning, and preheating processes simply by specifying relevant parameters with the docker run command. | China (Beijing) | 2023-04-11 | None |
| | China (Guangzhou) | 2023-04-10 | |
| | China (Shanghai) | 2023-04-07 | |
Node pools support modifying the operating system | The node pool supports modifying the operating system used by nodes, meeting the requirement to replace the operating system image in scenarios such as when the content of the operating system image changes. | China (Beijing) | 2023-04-11 | |
| | China (Guangzhou) | 2023-04-10 | |
| | China (Shanghai) | 2023-04-07 | |
Workloads support displaying image addresses | [Beta · Apply for Trial] The workload list page supports displaying the address of the container image. Allows users to view the image address, thereby assisting in decision-making for workload-related operations. | China (Beijing) | 2023-04-11 | |
| | China (Guangzhou) | 2023-04-10 | |
| | China (Shanghai) | 2023-04-07 | |
Deployment supports batch redeployment | [Beta · Apply for Trial] Deployment supports batch redeployment. Support multiple application scenarios for users and optimize the usage of the redeployment feature. | China (Beijing) | 2023-04-11 | |
| | China (Guangzhou) | 2023-04-10 | |
| | China (Shanghai) | 2023-04-07 | |
Optimize baseline inspection | Optimize the baseline inspection policy set for multiple Kubernetes versions. Ensure the baseline security of clusters with different Kubernetes versions | China (Beijing) | 2023-04-11 | |
| | China (Guangzhou) | 2023-04-10 | |
| | China (Shanghai) | 2023-04-07 | |
Node now supports certain GPU Elastic Compute Service specifications with A10 graphics cards | [Beta · Apply for Trial] The node now supports additional GPU Elastic Compute Service specifications with A10 graphics cards, including Compute-optimized gni2, Rendering-optimized gni2-vws, and other specifications. Meet users' business requirements for various GPU models. GPU Rendering instance family is a beta instance family in Elastic Compute Service (ECS). To use it, you need to submit an application in Elastic Compute Service (ECS). | China (Beijing) | 2023-04-11 | None |
| | China (Guangzhou) | 2023-04-10 | |
| | China (Shanghai) | 2023-04-07 | |
Upgrade the nvidia-device-plugin add-on | Upgrade the nvidia-device-plugin add-on version and align its features with those of the community NVIDIA device plugin for Kubernetes. | China (Beijing) | 2023-04-11 | |
| | China (Guangzhou) | 2023-04-10 | |
| | China (Shanghai) | 2023-04-07 | |
New dashboard for control plane add-on metrics in Prometheus monitoring | Beta: Prometheus monitoring adds a dashboard for control plane component metrics (API Server, ETCD, Scheduler), allowing users to observe the operational status of the cluster more intuitively. | China (Beijing) | 2023-04-11 | |
| | China (Guangzhou) | 2023-04-10 | |
| | China (Shanghai) | 2023-04-07 | |
Service supports configuring passthrough mode via annotations | A LoadBalancer-type Service supports configuring the LB-to-pod direct access mode through annotations. Expand the use cases for LB-to-pod direct access mode. | China (Beijing) | 2023-04-11 | |
| | China (Guangzhou) | 2023-04-10 | |
| | China (Shanghai) | 2023-04-07 | |
Optimize the container image selection experience | When creating a workload and selecting a container image, the system performs preliminary checks and provides guidance on relevant solutions to enhance the user experience. | China (Beijing) | 2023-04-11 | |
| | China (Guangzhou) | 2023-04-10 | |
| | China (Shanghai) | 2023-04-07 | |
March 2023
Feature name | Description | Region | Release date | Related documentation |
Feature name | Description | Region | Release date | Related documentation |
Storage supports capacity-type NAS | [beta] Dynamic and static PersistentVolumes support capacity-type NAS storage. They meet users' requirements for general file-sharing business scenarios. | China (Beijing) | 2023-03-28 | |
| | China (Guangzhou) | 2023-03-27 | |
| | China (Shanghai) | 2023-03-22 | |
By default, security hardening is enabled for node pools, and you can modify this setting. | When creating a cluster or node pool, the node pool's security hardening feature is enabled by default. After creation, you can modify this setting. After modifying the security hardening switch state, the change applies only to subsequently added nodes and does not affect existing nodes. | China (Beijing) | 2023-03-28 | |
| | China (Guangzhou) | 2023-03-27 | |
| | China (Shanghai) | 2023-03-22 | |
Nodes now support the GPU Accelerated g1vc specification | Nodes now support the following GPU accelerated g1vc specifications, which are suitable for deep learning, high-performance computing, and similar scenarios:
| China (Guangzhou) | 2023-03-27 | None |
Optimizing the console add-on management page | When an add-on is in an abnormal state or its version is outdated, a notification will be displayed on the cluster's Add-on Management page. This enhances users' ability to self-repair add-ons and reduces risks to their business caused by add-on malfunctions, outdated versions, and related issues. | China (Beijing) | 2023-03-28 | None |
| | China (Guangzhou) | 2023-03-27 | |
| | China (Shanghai) | 2023-03-22 | |
Check the account balance when adding Pay-as-you-go type nodes | When manually scaling out a cluster by adding Pay-as-you-go type nodes, the system checks the current account balance to prevent failures caused by insufficient funds. | China (Beijing) | 2023-03-28 | |
| | China (Guangzhou) | 2023-03-27 | |
| | China (Shanghai) | 2023-03-22 | |
Default node pool allows batch removal of nodes | The default node pool allows batch removal of nodes. Reduces the removal time for multiple nodes and improves the management efficiency of the default node pool. | China (Beijing) | 2023-03-28 | |
| | China (Guangzhou) | 2023-03-27 | |
| | China (Shanghai) | 2023-03-22 | |
Optimizing reserved PID for kubelet in cluster | Kubelet config reserves PIDs (process IDs), allocating a portion of PID resources for the system. This prevents business Pods from exhausting all PID resources and rendering the node unavailable. | China (Beijing) | 2023-03-28 | |
| | China (Guangzhou) | 2023-03-27 | |
| | China (Shanghai) | 2023-03-22 | |
Collection of logs from some cluster control plane add-ons to Torch Log Service is supported. | It supports enabling control plane add-on logs in a cluster and collecting them in Torch Log Service. This further enhances the observability of the Volcengine Kubernetes Engine control plane and meets additional user needs for log observation. | China (Beijing) | 2023-03-13 | |
| | China (Guangzhou) | 2023-03-10 | |
| | China (Shanghai) | 2023-03-09 | |
Upgrading the log-collector add-on and container log metadata | Upgrade the log-collector add-on to unify Volcengine Kubernetes Engine Logs metadata with Torch Log Service, enabling users to make more comprehensive use of structured log collection capabilities. | China (Beijing) | 2023-03-13 | |
| | China (Guangzhou) | 2023-03-10 | |
| | China (Shanghai) | 2023-03-09 | |
The cluster supports creating cloud disk snapshots and restoring data using cloud disk snapshots | The cluster supports creating cloud disk snapshots and restoring data from cloud disk snapshots. It enables users to back up and recover application data. It further enhances the cluster's disaster recovery capabilities. The cloud disk snapshot feature is a beta feature on the Elastic Compute Service side. To use this feature, please submit an application on the ECS side. | China (Beijing) | 2023-03-13 | None |
| | China (Guangzhou) | 2023-03-10 | |
| | China (Shanghai) | 2023-03-09 | |
When the status of a key resource is in the failed state, the corresponding event can be reported to the Cloud Monitor event center. | [Beta · Apply for Trial] When the status of a critical resource in a container cluster is failed, the corresponding event can be reported to the Cloud Monitor event center. Through the event-driven mechanism, critical events are promptly delivered to users, enabling them to quickly become aware of these events and perform operation and maintenance tasks, thereby improving the observability of the container cluster. | China (Beijing) | 2023-03-13 | None |
| | China (Guangzhou) | 2023-03-10 | |
| | China (Shanghai) | 2023-03-09 | |
Nodes now support two instance families: Local-SSD i2g and Big Data d2s | Node specifications in Volcengine Kubernetes Engine now support two Elastic Compute Service (ECS) instance families: Local-SSD i2g and Big Data d2s. It meets the diverse usage requirements of users for different ECS specifications across various scenarios. | China (Beijing) | 2023-03-13 | None |
| | China (Guangzhou) | 2023-03-10 | |
| | China (Shanghai) | 2023-03-09 | |
Supporting the cluster audit feature | Cluster auditing supports the collection and storage of every API Server request event in Torch Log Service. When a major event occurs in the cluster, administrators can trace back the time and cause of the major event by reviewing the cluster audit logs. It can meet the security and compliance requirements of the service, and has further enhanced the observability of the cluster. | China (Beijing) | 2023-03-06 | |
| | China (Guangzhou) | 2023-03-06 | |
| | China (Shanghai) | 2023-03-06 | |
February 2023
Feature name | Description | Region | Release date | Related documentation |
Feature name | Description | Region | Release date | Related documentation |
Support for Kubernetes version 1.24 | (public beta) Supports selecting Kubernetes 1.24 when creating a cluster. Meets users' needs to keep up with native community releases. | China (Beijing) | 2023-02-27 | |
| | China (Guangzhou) | 2023-02-23 | |
| | China (Shanghai) | 2023-02-22 | |
VPC-CNI cluster supports adding Pod subnets | The network model is a VPC-CNI cluster and, based on existing subnets, supports adding Pod subnets and worker node subnets to meet the need for additional IP resources. | China (Beijing) | 2023-02-27 | |
| | China (Guangzhou) | 2023-02-23 | |
| | China (Shanghai) | 2023-02-22 | |
Node pool supports subnet scheduling priority | By configuring a multi-subnet priority scheduling policy, nodes are scheduled for scale-out based on subnet priority, making full use of subnet IP resources and improving the success rate of node scale-out. | China (Beijing) | 2023-02-27 | |
| | China (Guangzhou) | 2023-02-23 | |
| | China (Shanghai) | 2023-02-22 | |
Support custom images built with Tailor | [Beta · Apply for Trial] Supports building custom images using Tailor, the custom image building tool provided by Volcano Engine. Ensure that custom images meet the operational standards of the Volcano Engine environment. | China (Beijing) | 2023-02-27 | None |
| | China (Guangzhou) | 2023-02-23 | |
| | China (Shanghai) | 2023-02-22 | |
Flannel modular management | The Flannel network add-on is incorporated into standardized add-on management, making it easier for users to independently upgrade add-ons, change cluster subnet configurations, and perform other operations, while providing enhanced configurability. | China (Beijing) | 2023-02-27 | |
| | China (Guangzhou) | 2023-02-23 | |
| | China (Shanghai) | 2023-02-22 | |
Supporting load-aware scheduling | [Beta · Apply for Trial] Scheduling is performed based on the Pod Request and the actual resource load of the node. Avoid resource overload or idleness caused by discrepancies between Pod Request values and actual resource usage. | China (Beijing) | 2023-02-27 | |
| | China (Guangzhou) | 2023-02-23 | |
| | China (Shanghai) | 2023-02-22 | |
Optimizing the policy set for application inspection across multiple Kubernetes versions | Optimize the application inspection policy set for multiple versions of Kubernetes. Ensure normal operation of application inspection on Kubernetes versions 1.20 and 1.24. | China (Beijing) | 2023-02-27 | |
| | China (Guangzhou) | 2023-02-23 | |
| | China (Shanghai) | 2023-02-22 | |
Node images support the CentOS-compatible version of veLinux | Beta: Node operating system supports the veLinux CentOS-compatible image. It meets CentOS users' image compatibility requirements, eliminating the need to modify existing tools and reducing migration costs. | China (Beijing) | 2023-02-09 | |
| | China (Guangzhou) | 2023-02-08 | |
| | China (Shanghai) | 2023-02-07 | |
Node pool supports preemptible instances | [Beta · Apply for Trial] Added support for node pools with the preemptible instance payment type, and introduced features for preemptible instance compensation and for supplementing preemptible capacity with pay-as-you-go instances. Supports the node distribution ratio policy. It aims to maintain the specified ratio of on-demand and preemptible instances within the node pool. | China (Beijing) | 2023-02-09 | |
| | China (Guangzhou) | 2023-02-08 | |
| | China (Shanghai) | 2023-02-07 | |
Elastic resources support setting a priority scheduling policy | [Beta] By configuring the ResourcePolicy scheduling policy, workloads are scheduled according to the policy, reducing resource usage costs. | China (Beijing) | 2023-02-09 | |
| | China (Guangzhou) | 2023-02-08 | |
| | China (Shanghai) | 2023-02-07 | |
The csi-ebs add-on supports deployment to VCI. | The cluster uses the VPC-CNI network model and supports deploying the csi-ebs storage service add-on to Volcengine Container Instance (VCI). It meets the deployment requirements for CSI add-ons in Serverless scenarios. | China (Beijing) | 2023-02-09 | |
| | China (Guangzhou) | 2023-02-08 | |
| | China (Shanghai) | 2023-02-07 | |
The cr-credential-controller add-on supports deployment to VCI. | Clusters using the VPC-CNI container network model support deploying the CR-credential-controller component for password-free container image access to Volcengine Container Instance (VCI). It meets the requirement for pulling images from a Container Registry (CR) without credentials in Serverless Kubernetes scenarios. | China (Beijing) | 2023-02-09 | |
| | China (Guangzhou) | 2023-02-08 | |
| | China (Shanghai) | 2023-02-07 | |
Unified managed and unmanaged modes of the cluster-autoscaler add-on | After the cluster-autoscaler add-on is unified to support both managed and unmanaged modes, legacy unmanaged mode add-ons can support self-upgrade. | China (Beijing) | 2023-02-09 | |
| | China (Guangzhou) | 2023-02-08 | |
| | China (Shanghai) | 2023-02-07 | |
Enabling the ALB Ingress feature | The Application Load Balancer (ALB) type Ingress routing feature has completed its beta phase and is now fully available to all users. ALB provides exceptional business processing performance for Layer 7 application loads. | China (Beijing) | 2023-02-09 | |
| | China (Guangzhou) | 2023-02-08 | |
| | China (Shanghai) | 2023-02-07 | |
Nodes now support ecs.d2.4xlarge and ecs.d2.7xlarge instance types | Node specifications in Volcengine Kubernetes Engine now support and are compatible with certain Elastic Compute Service instance types, including ecs.d2.4xlarge and ecs.d2.7xlarge, meeting the usage requirements of different users for various ECS specifications in different scenarios. | China (Beijing) | 2023-02-09 | None |
| | China (Guangzhou) | 2023-02-08 | |
| | China (Shanghai) | 2023-02-07 | |
Application center supports OCI Charts pushed by Helm version 3.7 and above | The OCI format in Helm was inconsistent before and after version 3.7. The Volcengine Kubernetes Engine application center now supports OCI Charts pushed by Helm version 3.7 and later, meeting users' needs to push multiple versions of Charts | China (Beijing) | 2023-02-09 | |
| | China (Guangzhou) | 2023-02-08 | |
| | China (Shanghai) | 2023-02-07 | |
January 2023
Feature name | Description | Region | Release date | Related documentation |
Feature name | Description | Region | Release date | Related documentation |
VPC-CNI add-on management | The VPC-CNI network add-on is incorporated into standardized add-on management. This simplifies independent upgrades, cluster subnet configuration changes, and other operations, while offering greater configurability. | China (Beijing) | 2023-01-12 | |
| | China (Guangzhou) | 2023-01-11 | |
| | China (Shanghai) | 2023-01-10 | |
Users are not allowed to delete the CLB created by the cluster for the API Server | Users are prohibited from deleting CLB associated with the cluster API Server, as accidental deletion of these CLB resources could render your Volcengine Kubernetes Engine (VKE) cluster unavailable. | China (Beijing) | 2023-01-12 | None |
| | China (Guangzhou) | 2023-01-11 | |
| | China (Shanghai) | 2023-01-10 | |
Implement large-scale scheduling prediction optimization for Pending Pods | Cluster Autoscaler implements scheduling prediction optimizations for large-scale pending pod scenarios. These optimizations improve the stability and scalability of CA in these environments. They also alleviate scheduling prediction timeouts that would otherwise block normal scale-out. | China (Beijing) | 2023-01-12 | |
| | China (Guangzhou) | 2023-01-11 | |
| | China (Shanghai) | 2023-01-10 | |
Improvements to the console node list page, label page, and node removal logic | Optimized the console node list page, label page, and node removal logic to further enhance the user experience. | China (Beijing) | 2023-01-12 | Node |
| | China (Guangzhou) | 2023-01-11 | |
| | China (Shanghai) | 2023-01-10 | |
mGPU device add-on supports a health check mechanism | [beta] When the mGPU driver fails, Device Plugins can promptly detect the health status of the mGPU and remove the node from the schedulable resource list. Avoid continuously scheduling subsequent computational Jobs to problematic nodes, which could result in a large number of Job failures. | China (Beijing) | 2023-01-12 | None |
| | China (Guangzhou) | 2023-01-11 | |
| | China (Shanghai) | 2023-01-10 | |
Last updated: 2026.10.16 17:40:20
All rights reserved © Beijing Volcano Engine Technology Co., Ltd2024
