ZStack ZSphere 5.2.0 Release Notes

Highlights

The main highlights of ZStack ZSphere 5.2.0 are as follows:

• Expanded Virtualization Platform Capabilities

  This release adds host and VM monitoring metrics, custom Windows VMTools installation, and control over password changes through QGA. It also supports registering VMs from independent cloned storage copies on the original platform, viewing and managing incomplete VM snapshots, and joining the Experience Improvement Program.

New Features

This release includes the following new features.

• Additional Host and VM Monitoring Metrics

  New metrics include CPU Ready Time, CPU Readiness, I/O latency, and NIC receive and transmit packet drop counts. These metrics help analyze CPU scheduling contention, disk response, and network issues. Depending on the resource type, you can view aggregate or device-level data and configure resource alarms for the applicable metrics.

• Custom Windows VMTools Installation and QGA Password Change Control

  The Windows VMTools installation wizard lets you select Cloudbase-Init and the monitoring agent (ZWatch) as needed. VirtIO drivers and QEMU Guest Agent (QGA) are required components. During installation, you can also control whether the platform is allowed to change the VM password online through QGA.

• Register VMs from Cloned Storage Copies on the Original Platform

  You can connect an independent cloned storage copy to the original platform and register VMs using the VM configuration files and disk data in that copy. When registering cloned SAN storage, you can reset the storage UUID. If the VM UUID in the selected configuration file already exists, a new VM UUID is generated during registration.

• View and Manage Incomplete VM Snapshots

  VM snapshots display their completeness status. If some disk snapshots are missing while others remain, the VM snapshot is shown as Incomplete, and you can still view the remaining disk snapshots. Incomplete snapshots do not support Full Recovery. If the VM snapshot does not contain a memory snapshot, you can use Custom Recovery to select disks available for recovery. Deleting an incomplete VM snapshot also removes its remaining disk snapshots.

• Experience Improvement Program

  Participation in the Experience Improvement Program is voluntary, and you can join or leave at any time. After you join, the platform collects technical data about product usage and runtime performance to help improve usability, performance, and stability.

Improvements

This release includes the following improvements.

• Theme Mode Settings

  Light Mode, Dark Mode, and System are supported. Selecting System automatically switches the platform interface between light and dark modes based on the operating system settings.

• Changes to Management Node HA Upgrade Behavior

  When running zsha2 upgrade-ha, the gencfg parameter now defaults to true instead of false. By default, the command updates database configurations and restarts MariaDB on both nodes. You no longer need to specify -gencfg=true.

  If you explicitly specify -gencfg=false, database configuration updates and the corresponding MariaDB restarts are skipped. Fixes that depend on these configuration updates do not take effect.

  Note: Perform the upgrade during a maintenance window. You must stop the management services on both nodes before the upgrade; otherwise, the upgrade cannot proceed. Start the management services on both nodes after the platform upgrade succeeds.

Resolved Issues

This release resolves the following issues.

VMs and VMTools

• Fixes an issue where inaccurate memory accounting on the destination host during VM hot migration might cause multiple large-memory VMs to be placed on the same host and exhaust its memory.
• Fixes issues where Dynamic Resource Scheduling (DRS) schedules migrations for VMs attached to non-migratable PCI/GPU passthrough devices and might incorrectly release source device bindings when migration fails.
• Fixes an issue where VM configuration and GPU/PCI device allocations are not cleaned up after a VM is deleted.
• Fixes an issue where the configured number of QXL displays and boot order do not take effect on the first boot when a VM is started immediately after creation.
• Fixes an issue where rebooting a VM might return an error.
• Fixes an issue where collecting NVIDIA GPU monitoring data through Windows VMTools might cause excessive VM CPU usage. Update VMTools or the monitoring agent inside the VM and restart the monitoring service for the fix to take effect.
• Fixes an issue where the hostname does not take effect when a Windows VM is created with both a hostname and initialization content such as User Data or an initial password. Upgrading does not automatically reset the hostname of a VM that has already completed initialization.
• Fixes an issue where the platform incorrectly reports that the monitoring agent (ZWatch) is running on a Windows VM when it is not installed or not running.
• Fixes an issue where the selected run position is not displayed correctly when converting a template to a VM.

Hosts and High Availability

• Fixes an issue where HA detection on SAN storage might produce incorrect results under certain network and lock service failure conditions.
• Fixes an issue where reconnecting a host that runs a large number of VMs might time out when the reconnection process restarts libvirtd.
• Fixes an issue where host agent startup or restart might not complete because version queries or lock checks are blocked.
• Fixes an issue where abnormal IPMI command execution causes prolonged waits during host hardware information collection or monitoring startup. Also fixes an issue where errors from failed host identification operations are not reported promptly.
• Fixes an issue where changes to the cluster migration network configuration take effect only after hosts are reconnected.
• Fixes an issue where reconnecting a host does not synchronize a management network netmask changed in the host operating system, causing the platform to display and apply the old netmask in subsequent configurations. Reconnect affected hosts after upgrading.

Storage and Snapshots

• Fixes an issue where available capacity is calculated incorrectly after snapshots on SAN storage are merged and deleted.
• Fixes an issue where the details overview still displays an error after the current VM snapshot is deleted.
• Fixes an issue where automatic refresh of the snapshot list closes an open action menu.
• Fixes an issue where regular users cannot view details of VM snapshots created by an administrator for their VMs.
• Fixes an issue where the VM details page becomes abnormal when a regular user views a VM snapshot that includes data disks owned by another account.
• Fixes an issue where residual disk files are not cleaned up after registered VMs and data volumes on local or NFS storage are permanently deleted.

Backup and Restore

• Fixes an issue where deleting or permanently deleting a VM might fail if CBT backup export resources have not been released.
• Fixes an issue where disabling an already stopped CBT backup task again might return an error.
• Improves IPv4 CIDR validation for the CBT backup network to prevent invalid network configurations from causing backup task startup failures.
• Fixes an issue where the resource selection list is empty when configuring a backup storage alarm. Also improves associated resource information when editing and viewing alarms.
• Improves snapshot data cleanup when uninstalling API Proxy.
• Adjusts API Proxy logging to write only to api-proxy.log, avoiding duplicate entries in the system messages log.
• Fixes an issue where a snapshot deletion on SAN storage that takes longer than 60 seconds is incorrectly reported as failed during Commvault backup, causing the backup to fail.
• Fixes issues where the CPU socket count and cores per socket differ from the original configuration after a VM is restored through Veeam, and changing the cores per socket might prevent the restored VM from starting. Create a new backup after updating API Proxy. Configurations in existing backup points and already restored VMs are not corrected automatically.
• Fixes an issue where temporary data is not cleaned up after a successful Veeam backup, causing database backup files to continue growing. Historical backup session data requires separate cleanup.
• Fixes an issue where redeploying API Proxy in an HA environment might cause a Manager Unavailable error due to node certificate issues.
• Fixes an issue where data agent registration fails during Veeam incremental backup of a data disk because a large number of data extents makes the backup descriptor too large.
• Fixes an issue where the boot mode and NIC type differ from the original configuration after a VM is restored through Commvault. Generate a new backup point after updating API Proxy. Configurations in existing backup points are not corrected automatically.

Platform O&M

• Fixes an issue where the old service remains running and occupies a port after the management node operations service is upgraded, causing the new service to restart repeatedly.
• Fixes an issue where concurrently starting management node services might create multiple management service processes and cause startup to fail.
• Fixes a potential file descriptor leak during management service operation.
• Fixes an issue where a transient query failure when collecting RAID physical disk information through StorCLI might cause false disk removal or insertion alarms.
• Fixes an issue where two-factor authentication QR codes viewed or downloaded from the user details page are invalid.
• Fixes an issue where the description overlaps the security code input field in the two-factor authentication security code dialog.
• Fixes an issue where the page reports failure even though resources have been successfully shared with a user group.
• Fixes an issue where the operation log page becomes abnormal when some fields are missing from an operation log.
• Fixes an issue where the configuration page becomes abnormal when modifying platform SNMP parameters.
• Fixes an issue where viewing the host list of a cluster might display an error page due to abnormal uptime information.
• Fixes an issue where resource type names in the data center resource topology remain in Chinese after switching to the English UI.

Security Updates

This release includes the following security updates.

• Updates the Linux kernel packages in the Helix 8.4 (x86_64) software repository to fix CVE-2026-64561 and CVE-2026-74684. After updating the kernel, restart the system during a maintenance window and confirm that the new kernel is running for the fixes to take effect.
• Updates the Microsoft Visual C++ x64 runtime bundled with Windows VMTools to 14.44.35211.0 to fix CVE-2024-43590. The runtime must be updated separately on existing Windows VMs.