ZStack Cloud 5.5.38 supports reclaiming unused disk space after data deletion by using the q35 virtual motherboard and Virtio BLK Discard mechanism in VM instances. This reduces invalid occupancy and improves storage utilization. New eligible VM instances created through the UI use the q35 virtual motherboard by default. For existing VM instances, you can switch to the q35 virtual motherboard in Advanced Settings on the details page when the instance is stopped.

Note: To use this capability, make sure that the following requirements are met:

• The cloud platform runs in an H84R environment.
• The primary storage type is local storage or Ceph (ZStone distributed block storage).
• The VM instance architecture is x86_64; the operating system is Debian 12, Ubuntu 22.04.5, or Ubuntu 24.04.1; and the operating system kernel, Virtio driver, and file system support the Virtio BLK Discard mechanism.
• Before changing the motherboard type of an existing VM instance, back up data and check /dev/vdX, PCI addresses, passthrough devices, and special device topology configurations.

ZStack Cloud 5.5.38 provides the following load balancer improvements:

ZStack Cloud 5.5.38 supports setting time policies for security groups and VPC firewall rule sets. You can configure effective dates and time periods for rules to run once or repeat weekly, allowing security group rules and rules synchronized to associated firewalls to automatically take effect or stop taking effect as scheduled. This supports scenarios such as access during working hours, temporary authorization, and periodic access control, reducing manual enablement and disablement operations and the risk of omissions.

Each security group or rule set can have multiple time policies, one of which can be selected as the current policy. Rules are not time limited when no current policy is set. Time policies support local time and UTC and provide status display and expiry reminders.

ZStack Cloud 5.5.38 improves IPv4/IPv6 dual-stack deployment and access for the management network. Management nodes can be configured with both IPv4 and IPv6 addresses, and you can use either type of address to access the platform UI and API. Other nodes can access the platform through IPv4 or IPv6.

ZStack Cloud 5.5.38 adds CBT support for ZBS (CBD protocol) primary storage. You can perform full backups, incremental backups, and data recovery for root volumes and data volumes of VM instances. This feature requires ZBS 1.9.0.

ZStack Cloud 5.5.38 supports configuring multiple image cache pools in the same Ceph primary storage. You can add existing physical storage pools as image cache pools and select the image cache pool policy in global settings. The platform selects a cache according to the policy when you create a VM instance, change an image, or reset a root volume, improving image-use flexibility in multi-storage-pool scenarios.

ZStack Cloud 5.5.38 adds the VM Instance Entered Running State and VM Instance Entered Stopped State alarm rules. You can use these rules to create event alarms and track VM instance state changes.

ZStack Cloud 5.5.38 adds the Automatically Acknowledge Alarm Messages After Resource Alarm Recovery global setting. When enabled, related alarm and recovery messages are automatically acknowledged after a resource alarm meets the recovery condition, without manual intervention.

ZStack Cloud 5.5.38 adds an SMS gateway plugin type. You can use custom plugins to flexibly integrate various SMS gateway servers and send platform alarm messages by SMS.

ZStack Cloud 5.5.38 adds SNMP Trap v2c and v3 support in addition to existing v1 compatibility, meeting the integration and security requirements of more monitoring platforms.

ZStack Cloud 5.5.38 enhances configuration and O&M capabilities for DPU elastic baremetal instances. You can create instances by offering, configure multiple business networks, IP addresses, and data volumes, and perform O&M operations including start, stop, restart, delete, system reinstallation, and login password changes. You can also perform O&M through the VNC console and identify and use existing management network bond configurations on the DPU SoC.

ZStack Cloud 5.5.38 supports adding tags to baremetal nodes for resource filtering and classification.

ZStack Cloud 5.5.38 improves the display of VM instance migration task details. When you view a migration task, you can directly obtain key information about the VM instance, source host, target host, source primary storage, and target primary storage to quickly confirm the migration scope and execution status.

ZStack Cloud 5.5.38 supports installing and using the ZMigrate migration service through the App Center to migrate VM instances from other platforms to the current platform. VMware source platforms are currently supported.

Fixes an issue where allocated management network IP addresses were not released after subsequent network address allocation failed during VPC vRouter creation, preventing continued address occupancy.

Fixes an issue where a host could become disconnected when a VM instance configured with SR-IOV NIC passthrough started on a Hygon host, improving startup stability for SR-IOV VM instances on Hygon hosts.

Fixes an issue where duplicate management of the Vhost service could cause association failures when ZStone primary storage was connected through the Vhost protocol. After the fix, ZStone deployment manages the Vhost service centrally. Cloud Platform 5.5.38 must be used with ZStone 5.5.12.

Fixes an issue where lock exceptions on 512e sector devices in SharedBlock primary storage caused volume creation, attachment, or expansion to fail, improving storage lock recovery reliability.

Fixes an issue where resources reserved on the target host during VM instance live migration were not correctly included in capacity statistics, preventing the reserved resources from being used by other tasks.

Fixes an issue where a high availability protection service could be incorrectly determined to be available before it was ready when the Ceph primary storage network failed, preventing related high availability operations from continuing when protection conditions are insufficient.

Optimizes target network checks before online backup to prevent backup target network connection exceptions from affecting running VM instances.

Fixes an issue where duplicate uploads of shared parent images caused concurrent full backup tasks to fail, improving their stability.

Fixes an issue where backup data restoration tasks could fail during concurrent ImageStore export, export cancellation, or image reading, improving data protection task stability.

Updates the operating system kernel to fix the following vulnerabilities: CVE-2026-74684, CVE-2026-43499, CVE-2026-64561, CVE-2026-64600, CNNVD-2025-88283257, CNNVD-2025-25811155, CNNVD-2025-61091425, CNNVD-2025-93979693, and CNNVD-2025-99729243.