XenServer

Known issues

This article contains advisories and minor issues in the XenServer 8 release and any workarounds that you can apply.

General

  • The backup and restore capability in xsconsole is temporarily not available. We are actively investigating a replacement solution. (CP-48776)

  • When attempting to use the serial console to connect to a XenServer host, the serial console might refuse to accept keyboard input. If you wait until after the console refreshes twice, the console then accepts keyboard input. (CA-311613)

  • When read caching is enabled, it is slower to read from the parent snapshot than from the leaf. (CP-32853)

  • When attempting to log in to the dom0 console with an incorrect password, you receive the following error message: When trying to update a password, this return status indicates that the value provided as the current password is not correct. This error message is expected even though it relates to a password change, not a login. Try to log in with the correct password. (CA-356441)

  • If a XenServer host is powered off unexpectedly and restarted, when it attempts to recover VMs that had an attached vTPM, the vTPM can sometimes be missing from the VM. (CA-379928)

Graphics

  • When NVIDIA T4 added in pass-through mode to a VM on some specific server hardware, that VM might not power on. (CA-360450)

  • You cannot install the NVIDIA Virtual GPU Manager supplemental pack as part of a fresh installation of your XenServer host. After you have installed your hosts, you can install NVIDIA Virtual GPU Manager by using the xe CLI or XenCenter. For more information, see Install the NVIDIA vGPU Manager for XenServer.

Guests

  • If you attempt to live migrate a VM with dynamic memory control enabled to a target host where resources such as memory are very constrained, the migration can sometimes fail. (CA-380607)

  • If your VM enters sleep mode, your mouse might become unresponsive although your keyboard remains functional. Rebooting the VM or disabling and enabling the USB controller through Device Manager restores mouse functionality. (CA-399746)

Windows guests

  • For domain-joined Windows 10 VMs (1903 and later) with FireEye Agent installed, repeated successful RDP connections can cause the VM to freeze with 100% CPU usage in ntoskrnl.exe. Perform a hard reboot on the VM to recover from this state. (CA-323760)

  • When you create a UEFI VM, the Windows installation requires a key press to start. If you do not press a key during the required period, the VM console switches to the UEFI shell.

    To work around this issue, you can restart the installation process in one of the following ways:

    • In the UEFI console, type the following commands.

       EFI:
       EFI\BOOT\BOOTX64
      
    • Reboot the VM

    When the installation process restarts, watch the VM console for the installation prompt. When the prompt appears, press any key. (CA-333694)

  • When attempting to update a Windows 10 VM from 1909 to 20H2 or later, the update might fail with a blue screen showing the error: INACCESSIBLE BOOT DEVICE. (XSI-1075)

    To make it less likely that this failure occurs, you can take the following steps before attempting to update:

    1. Update the XenServer VM Tools for Windows on your VM to the latest version.
    2. Snapshot the VM.
    3. In the VM registry, delete the following values from the HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\XENFILT\Parameters key: ActiveDeviceID, ActiveInstanceID, and ActiveLocationInformation
  • When creating a Windows VM from a template that is set to not automatically update its drivers, the created VM is incorrectly set to update its drivers. To work around this issue, run the following command: xe pool-param-set policy-no-vendor-device=true uuid=<pool-uuid>. This command ensures that future VMs created from the template are correctly set to not automatically update drivers. VMs that were previously generated from the template are not changed. (CA-371529)

  • If two disks are present during installation, Windows Server 2025 VMs might hang during boot. To avoid this issue, complete the installation with a single disk, then attach the second disk afterwards. (CA-399175)

  • If you boot a large number of Windows Server 2025 VMs on a single host, some might fail to start. (CP-51655)

For details on the latest XenServer VM Tools for Windows release, see What’s new.

Linux guests

  • You cannot use the Dynamic Memory Control (DMC) feature on Red Hat Enterprise Linux 8, Red Hat Enterprise Linux 9, Rocky Linux 8, Rocky Linux 9, or CentOS Stream 9 VMs as these operating systems do not support memory ballooning with the Xen hypervisor. (CA-378797)

  • On some Linux VMs, especially busy systems with outstanding disk I/O, attempts to suspend or live migrate the VM might fail. To work around this issue, try increasing the value of /sys/power/pm_freeze_timeout, for example, to 300000. If this work around is not successful, you can upgrade the Linux kernel of the VM to the latest version. (CP-41455)

  • If you install Debian 10 (Buster) by using PXE network boot, do not add console=tty0 to the boot parameters. This parameter can cause issues with the installation process. Use only console=hvc0 in the boot parameters. (CA-329015)

  • Due to a known issue with some SUSE Linux operating systems, if you attempt to trigger a crash dump on a SUSE Linux VM with 32 or more vCPUs, the operation fails and the VM does not restart automatically. This issue affects the following operating systems: SUSE Linux Enterprise Server 15 SP1, 15 SP2, 15 SP3, 15 SP4. (CA-375759)

  • If you create a VM with RHEL 8.7 or earlier and only 1 vCPU, the VM times out while booting. To work around this issue, use RHEL 8.8 or later or change the number of vCPUs to be 2 or more. (CA-376921)

  • Due to a known issue with Mesa drivers on Ubuntu 24.04, after installing an Ubuntu 24.04 (desktop) VM on an AMD platform host with the Wayland protocol enabled by default, you might experience repeated re-login prompts.

    We recommend to work around this issue by disabling Wayland or upgrading your Mesa drivers:

    • Disable Wayland

      Open a command prompt and run the following command:

       sudo nano /etc/gdm3/custom.conf
       <!--NeedCopy-->
      

      Set the following value to disable Wayland:

       WaylandEnable=false
       <!--NeedCopy-->
      

      Restart the VM to complete the process.

    • Update your Mesa drivers

      Run the following commands:

       sudo add-apt-repository ppa:kisak/kisak-mesa
       sudo apt update
       sudo apt upgrade
       <!--NeedCopy-->
      

      Restart the VM to complete the process.

Installation

  • When upgrading to or installing XenServer 8 from an ISO located on an IIS server, the install or upgrade can fail and leave your hosts unable to restart. The remote console shows the GRUB error: “File ‘/boot/grub/i3860pc/normal.mod’ not found. Entering rescue mode”. This issue is caused by the IIS configuration causing package files to be missing. To work around this issue, ensure that double escaping is allowed on IIS before extracting the installation ISO on it. (XSI-1063)

  • Use the latest XenCenter to upgrade from Citrix Hypervisor 8.2 CU1 to XenServer 8. Using an older version of XenCenter can result in a loss of connectivity.

    Download the latest XenCenter from the XenServer product downloads page.

Internationalization

  • Non-ASCII characters, for example, characters with accents, cannot be used in the host console. (CA-40845)

  • In a Windows VM with XenServer VM Tools for Windows installed, copy and paste of double-byte characters can fail if using the default desktop console in XenCenter. The pasted characters appear as question marks (?).

    To work around this issue, you can use the remote desktop console instead. (CA-281807)

SDK

  • In the Java SDK, some calls might fail with a JsonMappingException. (CA-393866)

Storage

  • If you create a new installation of XenServer on a host with a local XFS SR on an NVMe device, your local storage does not attach on boot. The action fails with the error: “Raised Server_error(SR_BACKEND_FAILURE, [ FileNotFoundError; [Errno 2] No such file or directory: ‘/sys/block/nvme0n/queue/scheduler’ ])”.

    After applying the latest updates, you can attach the local storage manually.

  • If you use GFS2 SRs and have two hosts in your clustered pool, your cluster can lose quorum and fence during an upgrade. To avoid this situation, either add a host to or remove a host from your cluster. Ensure that you have either one or three hosts in your pool during the upgrade process. (CA-313222)

  • If you are using a GFS2 SR and your cluster network is on a non-management VLAN, you cannot add or remove hosts to your clustered pool. (XSI-1604)

  • After removing an HBA LUN from a SAN, you might see log messages and I/O failures when querying Logical Volume information. To work around this issue, reboot the XenServer host. (XSI-984)

  • You cannot set or change the name of the tmpfs SR used by the PVS-Accelerator. When the type is tmpfs, the command xe sr-create disregards the value set for name-label and instead uses a fixed value. If you attempt to run the command xe sr-param-set to change the name of the tmpfs SR, you receive the error SCRIPT_MISSING.

  • You cannot run more than 200 PVS-Accelerator-enabled VMs on a XenServer host. (CP-39386)

  • A performance regression affecting writes to newly created virtual disk images (VDIs) on GFS2 and XFS storage repository (SR) types has been fixed.

    Note:

    This issue affects VDIs created after the software updates released to Early Access on 9 October 2024 and to Normal on 16 October 2024. VDIs created before these updates, or after the update published on 21 October 2024, are unaffected.

    While this update resolves the performance regression, it does not automatically restore performance for impacted VMs due to the structure of the backing file created during the regression period. To restore performance:

    • For non-persistent or layered VMs (managed with Citrix Virtual Apps and Desktops MCS): Perform a base image update.
    • For persistent VMs (including master images): Take a snapshot to create a new backing file for future writes, which resolves the performance regression. Keep the snapshot until a future XenServer update is released to prevent the issue from reoccurring. Deleting it prematurely could result in the previous incorrect metadata structure being reapplied, reintroducing the performance issue.
    • For test VMs: Recreate them to resolve the issue.

Third-party

  • A limitation in recent SSH clients means that SSH does not work for usernames that contain any of the following characters: {}[]|&. Ensure that your usernames and Active Directory server names do not contain any of these characters.

XenCenter

For information about known and fixed issues in XenCenter, see XenCenter What’s New.

Known issues