Release Date | Version Number | Change Content | Limitations and Impacts |
2026-09-21 | v1.1.22 | The CBS DescribeDisks API returns a contract message indicating insufficient slots. After CSI identifies the message, it reports the event to K8S Event to avoid invalid retries. Fixed an issue where nil panic occurred in CBS controller multi-replica scenarios because the EventRecorder of non-leader replicas was not initialized. | 1. Upgrading from version 1.1.6+ does not affect existing services. However, some components may be unavailable during the upgrade, so it is recommended to upgrade during off-peak hours. 2. When an upgrade from a version earlier than 1.1.6 is performed, the higher-version cbs component adds cpu and memory requests to the kube-system/csi-cbs-node daemonset, which may cause business pods to be evicted due to insufficient node resources. During the upgrade, you can modify the kube-system/csi-cbs-node daemonset: change the update policy to OnDelete, explicitly set the cpu and memory requests to 0, and then upgrade the cbs component in Component Management. |
2026-09-02 | v1.1.21 | Fixed the issue where snapshot-related CRD upgrade failed when the CBS component of TKE clusters earlier than version 1.26 was upgraded to v1.1.20. | 1. Upgrading from version 1.1.6+ does not affect existing services. However, some components may be unavailable during the upgrade, so it is recommended to upgrade during off-peak hours. 2. When an upgrade from a version earlier than 1.1.6 is performed, the higher-version cbs component adds cpu and memory requests to the kube-system/csi-cbs-node daemonset, which may cause business pods to be evicted due to insufficient node resources. During the upgrade, you can modify the kube-system/csi-cbs-node daemonset: change the update policy to OnDelete, explicitly set the cpu and memory requests to 0, and then upgrade the cbs component in Component Management. |
2026-08-26 | v1.1.20 | Added the snapshot.storage.k8s.io v1 CRD. Supports dynamic detection of the kubelet root-dir directory to resolve inconsistent kubelet root-dir configurations across different nodes. | 1. Upgrading from version 1.1.6+ does not affect existing services. However, some components may be unavailable during the upgrade, so it is recommended to upgrade during off-peak hours. 2. When an upgrade from a version earlier than 1.1.6 is performed, the higher-version cbs component adds cpu and memory requests to the kube-system/csi-cbs-node daemonset, which may cause business pods to be evicted due to insufficient node resources. During the upgrade, you can modify the kube-system/csi-cbs-node daemonset: change the update policy to OnDelete, explicitly set the cpu and memory requests to 0, and then upgrade the cbs component in Component Management. |
2026-08-13 | v1.1.19 | Fixed the CSI gRPC Server hang issue and implemented a closed-loop mechanism for automatic fault detection and restart. Optimized container probes to detect whether the gRPC Server is providing services normally. Supports automatic cloud disk scale-out policies to enable automatic scaling and improve Ops efficiency. Added support for ReadWriteOncePod to meet the restriction that a single PVC can be mounted by multiple Pods on a single node. | 1. Upgrading from version 1.1.6+ does not affect existing services. However, some components may be unavailable during the upgrade, so it is recommended to upgrade during off-peak hours. 2. When an upgrade from a version earlier than 1.1.6 is performed, the higher-version cbs component adds cpu and memory requests to the kube-system/csi-cbs-node daemonset, which may cause business pods to be evicted due to insufficient node resources. During the upgrade, you can modify the kube-system/csi-cbs-node daemonset: change the update policy to OnDelete, explicitly set the cpu and memory requests to 0, and then upgrade the cbs component in Component Management. 3. The automatic scale-out feature requires TKE cluster version 1.20+. |
2026-07-29 | v1.1.18 | Fixed the issue where the CBS DescribeInstance TencentCloud API did not return disk names, causing incorrect nodecount calculation. | 1. Upgrading from version 1.1.6+ does not affect existing services. However, some components may be unavailable during the upgrade, so it is recommended to upgrade during off-peak hours. 2. When an upgrade from a version earlier than 1.1.6 is performed, the higher-version cbs component adds cpu and memory requests to the kube-system/csi-cbs-node daemonset, which may cause business pods to be evicted due to insufficient node resources. During the upgrade, you can modify the kube-system/csi-cbs-node daemonset: change the update policy to OnDelete, explicitly set the cpu and memory requests to 0, and then upgrade the cbs component in Component Management. |
2026-07-07 | v1.1.17 | The InstallAddon/UpdateAddon APIs support configuring the kube-system/csi-cbs-node daemonset strict-mode startup parameters. | 1. Upgrading from version 1.1.6+ does not affect existing services. However, some components may be unavailable during the upgrade, so it is recommended to upgrade during off-peak hours. 2. When an upgrade from a version earlier than 1.1.6 is performed, the higher-version cbs component adds cpu and memory requests to the kube-system/csi-cbs-node daemonset, which may cause business pods to be evicted due to insufficient node resources. During the upgrade, you can modify the kube-system/csi-cbs-node daemonset: change the update policy to OnDelete, explicitly set the cpu and memory requests to 0, and then upgrade the cbs component in Component Management. |
2026-06-22 | v1.1.16 | Fixed the issue where upgrading the cbs csi plugin failed for clusters with versions earlier than 1.20. Fixed the number of disks that can be mounted by cbs csi. This feature is disabled by default and can be enabled by setting the strict-mode startup parameter of the kube-system/csi-cbs-node daemonset. | 1. Upgrading from version 1.1.6+ does not affect existing services. However, some components may be unavailable during the upgrade, so it is recommended to upgrade during off-peak hours. 2. When an upgrade from a version earlier than 1.1.6 is performed, the higher-version cbs component adds cpu and memory requests to the kube-system/csi-cbs-node daemonset, which may cause business pods to be evicted due to insufficient node resources. During the upgrade, you can modify the kube-system/csi-cbs-node daemonset: change the update policy to OnDelete, explicitly set the cpu and memory requests to 0, and then upgrade the cbs component in Component Management. |
2026-06-05 | v1.1.15 | Fixed the Event message that prompted a Pod restart during cbs disk scale-out. | 1. Upgrading from version 1.1.6+ does not affect existing services. However, some components may be unavailable during the upgrade, so it is recommended to upgrade during off-peak hours. 2. When an upgrade from a version earlier than 1.1.6 is performed, the higher-version cbs component adds cpu and memory requests to the kube-system/csi-cbs-node daemonset, which may cause business pods to be evicted due to insufficient node resources. During the upgrade, you can modify the kube-system/csi-cbs-node daemonset: change the update policy to OnDelete, explicitly set the cpu and memory requests to 0, and then upgrade the cbs component in Component Management. |
2026-04-14 | v1.1.14 | Added support for configuring the burstPerformance parameter on general-purpose SSD disks to enable SSD+burst IO capability. | 1. Upgrading from version 1.1.6+ does not affect existing services. However, some components may be unavailable during the upgrade, so it is recommended to upgrade during off-peak hours. 2. When an upgrade from a version earlier than 1.1.6 is performed, the higher-version cbs component adds cpu and memory requests to the kube-system/csi-cbs-node daemonset, which may cause business pods to be evicted due to insufficient node resources. During the upgrade, you can modify the kube-system/csi-cbs-node daemonset: change the update policy to OnDelete, explicitly set the cpu and memory requests to 0, and then upgrade the cbs component in Component Management. |
2026-03-16 | v1.1.13 | The csi-cbs-controller component supports multi-AZ high availability deployment. Fixed an issue where snapshot deletion remained stuck in the deleting state. Supports SSD+burst IO capability. You can configure the burstPerformance parameter to enable the SSD+burst IO capability. Supports automatically selecting a CDC cluster to create cloud disks based on the node tke.cloud.tencent.com/cdcid tag. | 1. Upgrading from version 1.1.6+ does not affect existing services. However, some components may be unavailable during the upgrade, so it is recommended to upgrade during off-peak hours. 2. When an upgrade from a version earlier than 1.1.6 is performed, the higher-version cbs component adds cpu and memory requests to the kube-system/csi-cbs-node daemonset, which may cause business pods to be evicted due to insufficient node resources. During the upgrade, you can modify the kube-system/csi-cbs-node daemonset: change the update policy to OnDelete, explicitly set the cpu and memory requests to 0, and then upgrade the cbs component in Component Management. |
2026-01-16 | v1.1.12 | Set the PVC finalizer before creating a disk to prevent the PVC from being deleted during disk creation, which could cause disk leakage. When deleting a PV, you can set the --detach-when-delete parameter for the csi-cbs-controller cbs-csi container to detach the disk in use before deleting it, preventing the PV and disk from remaining stuck in the deleting state. This parameter is disabled by default.Optimized the disk query logic by adding a caching mechanism for disk quota queries to reduce the frequency of CBS API calls. You can dynamically enable or disable the disk IO burst attribute by setting the PVC annotation tke.cloud.tencent.com/burst-performance-expected to "true" or "false".You can set the --set-encrypt-label parameter for the csi-cbs-node cbs-csi container to automatically detect the disk encryption types supported by nodes and add the tke.cloud.tencent.com/cbs-encrypt-type tag to nodes. This parameter is disabled by default.When cluster tags are synchronized to disks, Volumes that contain multiple disks are supported. Use TencentOS 4 as the base image to enhance component security. | 1. Upgrading from version 1.1.6+ does not affect existing services. However, some components may be unavailable during the upgrade, so it is recommended to upgrade during off-peak hours. 2. When an upgrade from a version earlier than 1.1.6 is performed, the higher-version cbs component adds cpu and memory requests to the kube-system/csi-cbs-node daemonset, which may cause business pods to be evicted due to insufficient node resources. During the upgrade, you can modify the kube-system/csi-cbs-node daemonset: change the update policy to OnDelete, explicitly set the cpu and memory requests to 0, and then upgrade the cbs component in Component Management. |
2025-06-13 | v1.1.11 | Enhance component security and upgrade the token acquisition method. When mounting a disk, check the mount path and disk type to avoid mounting a local directory. The csi-attacher container supports setting the kube-api-qps and kube-api-burst parameters to enable high-concurrency disk mounting. Increase the retry interval of the snapshot component to avoid sending too many requests during failed retries. The csi-cbs-controller cbs-csi container supports setting the kube-api-qps and kube-api-burst parameters to avoid rate limiting when creating and deleting a large number of snapshots. Added liveness health check. | 1. Upgrading from version 1.1.6+ does not affect existing services. However, some components may be unavailable during the upgrade, so it is recommended to upgrade during off-peak hours. 2. When an upgrade from a version earlier than 1.1.6 is performed, the higher-version cbs component adds cpu and memory requests to the kube-system/csi-cbs-node daemonset, which may cause business pods to be evicted due to insufficient node resources. During the upgrade, you can modify the kube-system/csi-cbs-node daemonset: change the update policy to OnDelete, explicitly set the cpu and memory requests to 0, and then upgrade the cbs component in Component Management. |
2025-03-21 | v1.1.10 | When creating a StorageClass, you can set the encryption type and key. When mounting/unmounting/scaling out a disk, check the current disk status to avoid sending invalid requests to the cbs backend. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2025-02-12 | v1.1.9 | Supported special disk types such as high IO. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2025-01-06 | v1.1.8 | During the upgrade, retain the customized CPU and memory resource configurations for containers. During the upgrade, retain the customized parameters such as kube-api-qps, kube-api-burst, worker-threads, resync-period, and ports. The image pull policy is changed from Always to IfNotPresent. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2024-09-26 | v1.1.7 | Supports local ultra-fast snapshots. Fixed the issue of norm token expiration. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2024-06-06 | v1.1.6 | Supports shared disks, allowing Pods to be mounted across multiple nodes. Supports setting request/limit for workloads. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2024-04-26 | v1.1.5 | Supported custom tags. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2023-12-04 | v1.1.4 | Supported native node CVM mode. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2023-11-21 | v1.1.3 | Optimization: Idempotency is supported when snapshots/cloud disks are deleted. A health check is added to driver-registrar. The plugin optimizes call time for scenarios such as creating, mounting, unmounting, and scaling out cloud disks. New feature: Intree PVs in clusters of earlier versions also support snapshot backup. The cbs plugin integrates with the resource management center and performs an audit before calling the resource deletion API. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2023-08-07 | v1.1.2 | csi-provisioner Supports fast data recovery for PVCs using snapshot IDs. csi-attacher Add a backoff algorithm to prevent high-frequency calls to the CBS attach/detach API. cbs-csi Supports CDC CBS scenarios. Optimize the time for Pods to mount disks during startup. You can set additional performance when creating an enhanced SSD. CBS snapshot idempotency optimization. CBS disk attachment is prohibited for CHC, bare metal, and registered nodes. Fixed the Bug where the LVM feature was unavailable in version 1.24. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2023-03-24 | v1.1.1 | Compatible with TKE 1.24. Compatible with legacy billing mode parameters. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2023-02-21 | v1.1.0 | Supports automatic detection of the maximum number of attachable disks for CVMs. Optimize the disk detachment logic to ensure that disk status remains normal. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2022-11-23 | v1.0.9 | Supports LVM striped mode. Increase the csi-resizer quota. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2022-11-15 | v1.0.8 | Supported LVM multi-disk management. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2022-10-20 | v1.0.7 | Supports snapshots for the xfs file system. Supports disk creation with 1G granularity. Supports block-type volumes. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2022-09-13 | v1.0.6 | Supports scaling out when a restore from a snapshot is performed. Supports scaling out the xfs file system. Supports adding super nodes to TKE clusters. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2022-07-06 | v1.0.5 | Supports disk creation when cloud tags are duplicated. Resolve the timing issue in online scaling out. Revoke permissions for the csi-node secret resource. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2022-05-31 | v1.0.4 | Optimize the plugin startup logic. The default concurrency of csi-attacher is increased to 50. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2022-05-06 | v1.0.3 | The plugin supports configuring taint tolerations. The plugin adds a type startup parameter. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2022-04-24 | v1.0.2 | Remove the directory cleanup logic from the NodeUnpublishVolume API of the plugin. The plugin supports obtaining the disk identifier by Serial. When the plugin is deleted, the corresponding crd resources are retained. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
2022-03-16 | v1.0.1 | Supported in-place lossless migration of workloads using intree cbs to csi when the cluster is upgraded from 1.18 to 1.20. | This upgrade will not affect existing services. Components may be unavailable during the upgrade. It is recommended to upgrade during off-peak hours. |
フィードバック