| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| How to Buy | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| XenSource : Documentation : | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Release 4.0.1Copyright © 2007 XenSource, Inc. Xen®, XenSource™, XenEnterprise™, XenServer™, XenExpress™, XenCenter™ and logos are either registered trademarks or trademarks of XenSource, Inc. in the United States and/or other countries. Other company or product names are for informational purposes only and may be trademarks of their respective owners. This product contains an embodiment of the following patent pending intellectual property of XenSource, Inc.:
October 2007 Table of Contents
Table of Contents This document is a guide to creating Virtual Machines with XenServer™, the platform virtualization solution from XenSource™. It describes the various methods of getting VMs up and running on XenServer Hosts for each of the supported operating systems. This section summarizes the rest of the guide so that you can find the information you need. The following topics are covered:
This document is primarily aimed at system administrators who need to set up deployments of XenServer VMs. Other documentation shipped with this release includes:
Table of Contents This chapter provides an overview of how VMs are created and lists virtual memory and virtual disk size minimums, describes the differences in virtual device support for the members of the XenServer product family. This chapter also discusses physical to virtual conversion (P2V), cloning Templates, and importing previosuly-exported VMs. VMs are created from Templates. A Template is a "gold image" that contains all the various configuration settings to instantiate a specific VM. XenServer ships with a base set of Templates, which range from generic "raw" VMs that can boot an OS vendor installation CD (Windows) or run an installation from a network repository (Red Hat Enterprise Linux, Suse Linux Enterprise 10) to complete pre-configured OS instances (Debian Etch and Sarge). There are three basic methods by which VMs are created using Templates:
Creating VMs by installing Windows operating systems onto the appropriate Templates is described in Chapter 3, Installing Windows VMs. Creating VMs by installing Linux operating systems onto the appropriate Templates is described in Chapter 4, Installing Linux VMs. Additionally, VMs can be created by These methods are describe in this chapter. In general, when installing VMs, be sure to follow the memory and disk space guidelines of the operating system and any relevant applications that you want to run when allocating resources such as memory and disk space.
The current version of the XenServer product family has the following general limitations on virtual devices for VMs. Note that specific guest operating systems may have lower limits for certain features. These limitations are noted in the individual guest installation section.
XenExpress, XenServer, and XenEnterprise also differ in the following ways that are relevant for creating VMs:
If you attempt to create a fifth VM with a XenExpress license, for example, an error message will appear, which suggests that you can upgrade your license and gives a web address for the XenSource website. Physical to Virtual Conversion (P2V) is the process by which an existing operating system on a physical server — its filesystem, configuration, etc. — is cast into a virtualized instance of the same operating system and filesystem, transferred, instantiated, and started as a VM on the XenServer Host. For existing physical instances of Windows servers, a third-party tool is required. Windows P2V software and documentation is available for download from the XenSource website at http://www.xensource.com/partners/offers. For existing physical instances of Linux servers, this is accomplished by booting from the XenServer installation CD and choosing the P2V option. The filesystem is copied across the network onto a XenServer Host, where it appears as a normal VM. We recommend that you perform P2V operations during off-peak hours since the process involves transferring a large amount of data, which could impact other Virtual Machines running on the XenServer Host. The P2V tool requires a 64-bit capable CPU by default. If you have an existing Linux instance on an older machine that you want to transfer via P2V, you can boot the CD via the p2v-legacy option at the initial prompt. This does require at least a PAE-enabled machine, so for very old machines you can physically move the hard drive to a PAE-enabled machine and perform the operation from there. To P2V an existing Linux server directly to a XenServer Host
When the P2V process is complete and the new VM is created, you will need to create and attach a VIF for it to have external network connectivity. Similarly, extra disks may also be added to take advantage of additional storage capacity available to the XenServer host. Since the VM has new virtual network hardware, the MAC addresses it sees will also be different. Follow the Linux cloning guidelines (see Section 4.5, “Preparing to clone a Linux VM”) for customizing the configuration files to make the VM re-run any hardware detection scripts at startup. When considering how to best begin virtualizing a collection of physical servers, it is best to gain some comfort level and experience with virtualizing servers that are more simply configured, moving later to servers with more complex configurations. Good candidates typically include servers that are used for test and development environments, and servers used for in-house IT infrastructure (intranet web servers, DNS, NIS, and other network services, etc.). Typically servers that are doing heavily CPU-intensive tasks (sophisticated mathematical modeling, video rendering) or are I/O-intensive (high-traffic commercial web sites, highly-used database servers, streaming audio/video servers) are not the best candidates for virtualization at the start. Once you've identified some physical servers that seem reasonable to work on first, you should take a close look at how you are currently using them. What applications are they hosting? How I/O intensive are they? How CPU-intensive are they? To make a reasonable assessment, you should gather a reasonable amount of data on the current physical servers that you are thinking about virtualizing. Look at system monitoring data for disk usage, CPU usage, memory usage, and network traffic, and consider both peak and average values. Good candidates for virtualization are:
Once you have identified a set of machines that you want to virtualize, you should plan the process to accomplish the task. First, provision the physical servers that will serve as your XenServer Hosts. The chief constraint on the number of VMs you can run per XenServer Host is system memory. Next, plan how you will create the VMs. Your choices are to P2V an existing server, install a fresh server from network-mounted vendor media, or install a base operating system using a pre-existing template. If you P2V an existing server, it's best to P2V a test instance of the server, and run it in parallel with the existing physical server until you are satisfied that everything works properly in the virtual environment before re-purposing the existing physical machine. Next, plan how to arrange the desired VMs on the XenServer Hosts. Don't "mix up" servers - assign VMs to specific XenServer Hosts, giving consideration to complementary resource consumption (mixing CPU-intensive and I/O-intensive workloads) and complementary peak usage patterns (for instance, assigning overnight batch processing and daytime interactive workloads to the same XenServer Host). For configuring individual VMs themselves, keep these guidelines in mind:
You can make a copy of an existing VM by cloning from a template. Templates are just ordinary VMs which are intended to be used as master copies to instantiate copies from. A VM can be customized and converted into a template, but be sure to follow the appropriate preparation procedure for the VM (see Section 3.4, “Preparing to clone a Windows VM” for Windows and Section 4.5, “Preparing to clone a Linux VM” for Linux). Templates cannot be used as normal VMs without first cloning them. XenServer has two mechanisms for cloning VMs: a full copy, or a faster "Copy-on-Write" (CoW) mode which only writes modified blocks to disk. The CoW mode is only supported for file-backed VMs. CoW is designed to save disk space and allow fast clones, but will slightly slow down normal disk performance. A template can be fast-cloned multiple times without slowdown, but if a template is cloned into a VM and the clone converted back into a template, disk performance can linearly decrease depending on the number of times this has happened. In this event, the vm-copy CLI command can be used to perform a full copy of the disks and restore expected levels of disk performance. You can make a VM by importing an existing exported VM. Like cloning, exporting and importing a VM is a means for creating additional VMs of a certain configuration. You might, for example, have a special-purpose server configuration that you use many times. Once you have set up a VM the way you want it, you can export it, and later import it, creating another copy of your specially-configured VM. This also provides a simple way to move a VM to another XenServer Host. When importing a VM, you can choose to preserve the MAC address on any virtual network interfaces associated with it. If you do choose to generate a new MAC address, be sure to follow the appropriate preparation procedure for the imported VM (see Section 3.4, “Preparing to clone a Windows VM” for Windows and Section 4.5, “Preparing to clone a Linux VM” for Linux). Importing an exported VM will take some time, and depends on the size of the VM and the speed and bandwidth of the network connection between the XenServer Host and XenCenter. NoteNote that an exported VM that originated on one XenServer Host might or might not be able to be resumed on a different XenServer Host. For example, if a Windows VM created on a XenServer Host with an Intel VT-enabled CPU is exported, then imported to a XenServer Host with an AMD-V CPU, it will not start. Table of Contents XenServer allows you to install Windows 2000 SP4, Windows Server 2003 (32-/64- bit), or Windows XP SP2 into a VM. Installing Windows VMs on XenServer Host requires hardware virtualization support (Intel VT or AMD-V). Installing a Windows VM can be broken down into two main steps:
Windows VMs are installed by cloning an appropriate Template from either XenCenter or the CLI. The Templates for individual guests have predefined platform flags set which define the configuration of the virtual hardware. For example, all Windows VMs are installed with the ACPI Hardware Abstraction Layer (HAL). If you subsequently change one of these VMs to have multiple virtual CPUs, Windows automatically switches the HAL to multi-processor mode. The list of pre-defined Windows templates are:
The Windows VM can be installed either from a install CD in a physical CD-ROM on the XenServer Host, or from an ISO image of your Windows media (see Appendix A, Creating ISO images for information on how to make an ISO image from a Windows install CD and make it available for use). To make an ISO library available to XenServer Hosts, create an external NFS or SMB/CIFS share directory. The NFS or SMB/CIFS server must be set to allow root access to the share. For NFS shares, this is accomplished by setting the Then either use XenCenter to attach the ISO library, or connect to the host console and type in:
xe-mount-iso-sr Additional arguments to the mount command may be passed in, for advanced use. If making a Windows SMB/CIFS share available to the XenServer host, either use XenCenter to make it available, or connect to the host console and type in: xe-mount-iso-sr The xe-mount-iso-sr //server1/myisos -t cifs -o username=johndoe/mydomain xe-mount-iso-sr //server2/iso_share -t smbfs -o username=alice After mounting the share, any ISOs in it should be available by name from the CD pulldown list in XenCenter, or as CD images from the CLI commands. The ISO should be attached to an appropriate Windows template:
In XenServer 3.2 and earlier, ISOs could be copied directly to the control domain into the To use local ISO storage from the control domain
WarningBe extremely careful with copying ISOs directly onto the control domain filesystem, as it has limited space available. A network share is a much safer mechanism for storing large numbers of ISO images. If the control domain does fill up, unpredictable behavior will result. The XenSource paravirtualizated network and SCSI drivers provide high performance I/O services without the overhead of traditional device emulation found in first-generation virtualization products. During the installation of a Windows operating system, Xen will use traditional device emulation to present a standard IDE controller and a standard network card to the Virtual Machine. This allows Windows to complete its installation using built-in drivers, but with reduced performance due to the overhead inherent in emulation of the controller drivers. After Windows is installed, you install the XenSource high-speed paravirtualized drivers. These are on an ISO available to the virtual CD-ROM drive of the Virtual Machine. These drivers replace the emulated devices and provide high-speed transport between Windows and the XenServer product family software. NoteWhile a Windows VM will function without them, performance is significantly hampered unless these drivers are installed. Running Windows VMs without these drivers is not supported. Some features, such as live relocation across physical hosts, will only work with the paravirtual drivers installed and active. The Windows paravirtualized drivers ISO can be attached to the VM by using the "Install Tools" menu in XenCenter, or by directly inserting the built-in The graphical console for Windows can be either a standard console via emulated graphics card, or an RDP connection. For Windows VMs, there is a button on the Console tab. Clicking it disables the standard graphical console, and switches to using Remote Desktop instead. The button will be greyed out if you do not have Remote Desktop enabled in the VM, and the paravirtualized drivers must be installed. To enable Remote Desktop on a Windows VM
You need to use the Windows utility sysprep to prepare a Windows VM for cloning. This is the only supported way to properly clone a Windows VM. Computers running Windows operating systems use a Security ID (SID) to uniquely identify themselves. When cloning a Windows VM, it is important to take steps to ensure the uniqueness of these Security IDs. Cloning an installation without taking the recommended system preparation steps can lead to duplicate SIDs and other problems. Because the SID identifies the computer or domain as well as the user, it is critical that it is unique. Refer to the Microsoft KnowledgeBase article 162001, "Do not disk duplicate installed versions of Windows," for more information. Sysprep modifies the local computer Security ID (SID)to make it unique to each computer. The Sysprep binaries are on the Windows product CDs in the Here are the overall steps you need to follow to clone Windows VMs:
NoteThe original, sysprepped VM (the "source" VM) should not be restarted again after the sysprep stage, and should be converted to a template immediately afterwards to prevent this. If the source VM is restarted, sysprep must be run on it again before it can be safely used to make additional clones. For more information on using sysprep, refer to the Microsoft TechNet page "Windows System Preparation Tool." There are many versions and variations of Windows with different levels of support for the features provided by XenServer. This section lists notes and errata for the known differences. Multiple VCPUs are exposed as CPU sockets to Windows guests, and are subject to the licensing limitations present in the guest. The number of CPUs present in the guest can be confirmed by checking the Device Manager. The number of CPUs actually being used by Windows can be seen in the Task Manager. The disk enumeration order in a Windows guest may differ from the order in which they were initially added. This is a behavioral artifact between the paravirtualized drivers and the PnP subsystem in Windows. For example, the first disk may show up as "Disk 1", the next disk hotplugged as "Disk 0", a subsequent disk as "Disk 2", and then upwards in the expected fashion. Windows XP does not support disks larger than 2TB (terabytes) in size. See this article in the Windows Hardware Developer Central website. Table of Contents
XenServer supports the installation of many Linux distributions into paravirtualized VMs. There are three installation mechanisms at present: complete distributions provided as built-in templates, Physical-to-Virtual (P2V) of an existing native install (see Section 2.4, “Physical to Virtual Conversion (P2V)”), and using the vendor media to perform a network installation. All use of Linux VMs requires the Linux Pack to be installed onto the XenServer Host. The supported Linux distributions are:
Distributions which use the same installation mechanism as Red Hat Enterprise Linux 5 (e.g. Fedora Core 6) may successfully install using the same template. However, use of distributions not present in the above list is unsupported. This is the simplest way of installing a VM. The template provided with XenServer can be used to directly create a VM running version 3.1 (Sarge) or 4.0 (Etch) of the Debian Linux distribution without need for vendor installation media and without performing a P2V conversion of an existing physical server. The VMs are instantiated by using the vm-install from the CLI, or by cloning the template using XenCenter. For example, using the CLI on Linux: # xe vm-install template=Debian\ Etch\ 4.0 new-name-label=ExampleVM f21cd819-5b7d-002d-7a1e-861a954e770 When the VM is first booted, it will prompt you for a root password, a VNC password (for graphical use), and a hostname. After values are entered for these, it will finish at a standard login prompt, ready for use. You will need to add a network interface if installed via the CLI. The XenServer guest installer allows you to install an operating system from a network-accessible copy of vendor media into a VM. In preparation for installing from vendor media, you need to make an exploded network repository of your vendor media (not ISO images), exported via NFS, HTTP or FTP accessible to the XenServer Host administration interface. See Appendix B, Setting Up a Red Hat Installation Server for information on how to copy a set of installation CDs to a network drive. For further information, see the section "Preparing for a Network Installation" in the Red Hat Enterprise Linux 4 Installation Guide for details. The network repository must be accessible from the Control Domain of the XenServer host, normally via the management interface. The URL should point to the base of the CD/DVD image on the network server, and be of the form:
The XenServer "New VM Wizard" provides an additional step for vendor-installable templates which prompts for the repository URL. When using the CLI, install the template as normal using vm-install and then set the other-config-install-repository key to the value of the URL. When the VM is subsequently started, it will begin the network installation process. NoteWhen installing a new Linux-based VM, it is important to fully finish the installation and reboot it before performing any other operations on it. This is analogous to not interrupting a Windows installation, which would leave you with a non-functional VM. Older Linux distributions such as Red Hat Linux Enterprise 3.6 do not support Xen directly, and are typically legacy installations which benefit from virtualization for the purposes of server consolidation or hardware upgrades. The XenServer P2V feature analyzes existing installations and converts them into VMs. When an installation is converted into a VM using P2V (see Section 2.4, “Physical to Virtual Conversion (P2V)”), the kernel used is also automatically switched to a Xen paravirtualized kernel. XenServer contains ports of the Red Hat Enterprise Linux 3/4 and SUSE Enterprise Linux 9 kernels to support the native Xen hypervisor interface directly. These kernels are present in the built-in During the installation of a VM via P2V, a special network is used to assign a temporary IP address to the VM to enable the installation to proceed. It is possible that the range of IP addresses used might conflict with real IP addresses already in use in your network. The default range of IP addresses is 192.168.128.1 to 192.168.128.254, and the default netmask is 255.255.255.0. To change the guest installer network values
Although all the supported Linux distributions are natively paravirtualized (and thus do not need special drivers for full performance), XenServer includes a guest agent which provides additional information about the VM to the host. This additional information includes:
It is important to install this agent and keep it up-to-date (see Chapter 5, Updating VMs) as you upgrade your XenServer host. To install the guest agent
NoteCD-ROM drives and ISOs attached to Linux Virtual Machines appear as /dev/xvdd instead of as /dev/cdrom as you might reasonably expect. This is because they are not "true" CD-ROM devices, but normal devices. When the CD is ejected by either XenCenter or the CLI, it hot-unplugs the device from the VM and the device disappears. This is different from Windows Virtual Machines, where the CD remains in the VM in an empty state. When a Linux VM is cloned, some virtual hardware parameters are changed in the new VM. The VM may need to be customized in order to be aware of these changes. For instructions for specific supported Linux distributions, please see Section 4.8, “Release Notes”. Of course, a cloned VM is another computer, and like any new computer in a network, it must have a unique name within the network domain it is part of. A cloned VM must have a unique IP address within the network domain it is part of. This is not a problem in general if DHCP is used to assign addresses; when the VM boots the DHCP server will assign it an IP address. If the cloned VM had a static IP address, the clone must be given an unused IP address before being booted. In some cases, the MAC address of a cloned VM's virtual network interface is recorded in the network configuration files. After the VM is cloned, the new cloned VM has a different MAC address. Therefore, when started, the network does not come up automatically. Some Linux distributions use udev rules to remember the MAC address of each network interface, and persist a name for that interface. This is intended so that the same physical NIC always maps to the same ethn interface, which is particularly useful with removable NICs (like laptops). But this behavior is problematic in the context of Virtual Machines. For example, if you configure two virtual NICs when you install a VM, and then shut it down and remove the first NIC, on reboot XenCenter shows just one NIC, but calls it eth0. Meanwhile the VM is deliberately forcing this to be eth1. The result is that networking doesn't work. If the VM uses persistent names, the best thing to do is to turn these rules off. If for some reason you do not want to turn persistent names off, be aware that you will need to reconfigure networking inside the VM in the usual way, and the information shown in XenCenter will be out of synch with reality. By default, the clocks in a Linux VM are synchronized to the clock running on the control domain, and cannot be independently changed. This mode is a convenient default, since only the control domain needs to be running the NTP service to keep accurate time across all VMs. Upon installation of a new Linux VM, make sure you change the time-zone from the default UTC to your local value (see Section 4.8, “Release Notes” for specific distribution instructions). Individual Linux VMs can be set to maintain independent times
With the exception of VMs based on the Debian templates, VMs might not be set up to support VNC by default. For example, if you P2V a server that does not have a VNC server installed, the resulting VM won't have VNC installed either. Before you can connect with the XenCenter graphical console, you need to ensure that the VNC server and an X display manager are installed on the VM and properly configured. This section describes the procedures for configuring VNC on each of the supported Linux operating system distributions to allow proper interactions with the XenCenter graphical console. CentOS-based VMs should use the instructions for the Red Hat-based VMs below, as they use the same base code to provide graphical VNC access. CentOS 4 is based on Red Hat Enterprise Linux 4, and CentOS 5 is based on Red Hat Enterprise Linux 5. NoteBefore setting up your Red Hat VMs for VNC, be sure that you have installed the Linux guest agent. See Section 4.4, “Installing the Linux guest agent” for details. In order to configure VNC on Red Hat VMs, you need to modify the GDM configuration. The GDM configuration is held in a file whose location varies depending on the version of Red Hat Linux you are using. Before modifying it, we first must determine the location of this configuration file; this file will then be modified in a number of subsequent procedures in this section. If you are using Red Hat Linux version 3 or 4 the GDM configuration file is If you are using Red Hat Linux version 5 the GDM configuration file is
Restart GDM for your change in configuration to take effect, by running /usr/sbin/gdm-restart. Note that, for Red Hat Linux, runlevel 5 is used for graphical startup. If your installation is configured to start up in runlevel 3, you will need to change this in order for the display manager to be started (and therefore to get access to a graphical console). Please refer to Section 4.7.4, “Checking runlevels” for further details. The firewall configuration by default does not allow VNC to traffic to go through. If you have a firewall between the VM and XenCenter, you need to allow traffic over the port that the VNC connection uses. By default, a VNC server listens for connections from a VNC viewer on TCP port 5900 + N, where N is the display number (usually just zero). So a VNC server setup for Display-0 will listen on TCP port 5900, Display-1 is TCP-5901, etc. Consult your firewall documentation to make sure these ports are open. You might want to further customize your firewall configuration if you want to use IP connection tracking or limit the initiation of connections to be from one side only. To customize Red Hat-based VMs firewall to open the VNC port
Alternatively, you can disable the firewall until the next reboot by using service iptables stop, or permanently by using chkconfig iptables off. This can of course expose additional services to the outside world and reduce the overall security of your VM. If, after connecting to a Virtual Machine with the graphical console, the screen resolution is mismatched (for example, the VM's display is too big to comfortably fit in the Graphical Console pane), you can control it by setting the VNC server's
NoteBefore setting up your SUSE Linux Enterprise Server VMs for VNC, be sure that you have installed the Linux guest agent. See Section 4.4, “Installing the Linux guest agent” for details. SLES has support for enabling “Remote Administration” as a configuration option in YaST. You can select to enable Remote Administration at install time, available on the screen of the SLES installer. This will allow you to connect an external VNC viewer to your guest to view the graphical console; the methodology for using the SLES remote administration feature is slightly different than that provided by XenCenter, but it is possible to modify the configuration files in your SUSE Linux VM such that it is integrated with the graphical console feature. Before making configuration changes, you should verify that you have a VNC server installed. SUSE ships the tightvnc server by default; this is a suitable VNC server, but you can also use the standard RealVNC distribution if you prefer. You can check that you have the tightvnc software installed by running the command: rpm -q tightvnc If Remote Administration was not enabled during installation of the SLES software, you can enable it as follows:
After enabling Remote Administration, you need to modify a configuration file if you want to allow XenCenter to connect, or else use a third party VNC client.
SUSE Linux uses runlevel 5 for graphical startup. If your remote desktop does not appear, verify that your VM is configured to start up in runlevel 5. Refer to Section 4.7.4, “Checking runlevels” for details. The firewall configuration by default does not allow VNC to traffic to go through. If you have a firewall between the VM and XenCenter, you need to allow traffic over the port that the VNC connection uses. By default, a VNC server listens for connections from a VNC viewer on TCP port 5900 + N, where N is the display number (usually just zero). So a VNC server setup for Display-0 will listen on TCP port 5900, Display-1 is TCP-5901, etc. Consult your firewall documentation to make sure these ports are open. You might want to further customize your firewall configuration if you want to use IP connection tracking or limit the initiation of connections to be from one side only. To customize SLES-based VMs firewall to open the VNC port
Alternatively, you can disable the firewall until the next reboot by using the rcSuSEfirewall2 stop command, or permanently by using YaST. This can of course expose additional services to the outside world and reduce the overall security of your VM.
If, after connecting to a Virtual Machine with the Graphical Console, the screen resolution is mismatched (for example, the VM's display is too big to comfortably fit in the Graphical Console pane), you can control it by setting the VNC server's
The built-in Debian Sarge and Etch templates come pre-configured with VNC setup and ready use. However, the default VNC configuration in Debian does not permit the root administration user to log in by default. To log in by VNC, you can either:
If you need to reset the VNC password, use the command vnc4passwd /etc/vncpass Most modern Linux distributions support Xen paravirtualization directly, but have different installation mechanisms and some kernel limitations. XenServer includes a custom Xen kernel for Debian guests installed via the built-in template. This kernel is a cut-down version of the dom0 kernel used in XenServer, and as such is the most heavily tested and reliable guest kernel available. After installation, the time-zone in a Debian VM defaults to UTC (see Section 4.6, “Time handling in Linux VMs”). It can be changed to your local value by using the tzconfig command. To prepare a Debian guest for cloning (see Section 4.5.3, “MAC address”), Ethernet name persistence must be disabled. For Debian Sarge VMs, name persistence is controlled through rm -f /etc/udev/rules.d/z45_persistent-net-generator.rules XenServer includes a custom port of the RHEL3.8 kernel with native Xen paravirtualized guest support. This kernel is installed during the P2V process for RHEL3.6-3.8 guests. Since the kernel is based on Linux 2.4, the following limitations apply:
XenServer includes a custom port of the RHEL 4.5 kernel with additional bug fixes and expanded Xen support. This kernel is used in the vendor installation templates of RHEL 4.1 and 4.4, but not in the RHEL 4.5 template (since RHEL4.5 is the first release with native Xen support). The issues below have been reported upstream to Red Hat and are already fixed in our kernel (which can be installed by using the
To prepare a RHEL4 guest for cloning (see Section 4.5.3, “MAC address”), edit XenServer uses the standard Red Hat kernel supplied with RHEL5 as the guest kernel. Any bugs found in this kernel are reported upstream to Red Hat, and are listed below:
When you install the XenServer Please refer to Section 4.8.3, “Red Hat Enterprise Linux 4” for the list of CentOS 4 release notes. Unlike RHEL4, CentOS includes a third-party updates mechanism known as yum. The Please refer to Section 4.8.4, “Red Hat Enterprise Linux 5” for the list of CentOS 5 release notes. XenServer includes a custom port of the SLES9 SP3 kernel with additional bug fixes and expanded support. This kernel is automatically installed during the P2V process for SLES9 guests. To prepare a SUSE Linux guest for cloning (see Section 4.5.3, “MAC address”), edit FORCE_PERSISTENT_NAMES=yes to FORCE_PERSISTENT_NAMES=no When you P2V a SLES 9 server, the networking configuration files that were present on the physical server will remain on the VM. You may wish to move these aside, or update them accordingly, when you add virtual interfaces to the VM. XenServer uses the standard Novell kernel supplied with SLES10 SP1 as the guest kernel. Any bugs found in this kernel are reported upstream to Novell and listed below:
When installing SLES10 VMs, you may experience some display corruption in the text-mode install. This is more pronounced when using SLES10, and is partially fixed in SLES10 Service Pack 1. In either case, pressing “Control-L” after the corruption happens will force a screen redraw, and improve the installation experience. Table of Contents This chapter discusses updating VMs with new Linux kernel revisions, applying Windows Service Packs, and updates to XenSource paravirtualized drivers and VM utilities. Upgrades to VMs are typically required when moving to a new version of XenServer. The following are current issues involving upgrading VMs running on XenServer to this version:
The paravirtualized drivers are present on the built-in When updating Windows 2000 SP2 VMs from XenServer 3.2 to the current version, you may occasionally see a dialog popup containing “the xennet.sys device driver could not locate the entry point __XenTrace in driver xevtchn.sys”. This warning message is harmless and installation will continue normally after clicking past the dialog. The Linux guest utilities can be updated by re-running the
The updates are posted online at: http://updates.xensource.com/XenServer/4.0.1/. For example, the RHEL 3.x kernel would be at: http://updates.xensource.com/XenServer/4.0.1/rhel3x/. This is of particular importance for RHEL 4.5 and CentOS 4.5, where you will get the upstream kernel by default, which has certain limitations (see Section 4.8, “Release Notes”). For yum-enabled distributions (CentOS 4 and 5, RHEL 5), For Debian, XenServer can use ISO images of CD-ROM or DVD-ROM disks as installation media and data sources for Windows or Linux VMs. This section describes how to make ISOs from CD/DVD media. On a Linux computer
On a Windows computer
Table of Contents This chapter explains how to set up a server as an installation server for Red Hat Linux. For a server to act as a Red Hat Linux network installation server, you need space on your server to copy the entire contents of each CD onto your server. This is typically the number of CDs or ISO images times 650MB. Ensure that the space you intend to use is formatted with your chosen filesystem and is mounted. You can check this space with the command: df -h
Next, you need to make your installation data available to other machines on the network. You can use NFS, HTTP, or FTP protocols. You can enable all three services on your server or any subset of the three. To install over NFS you need to meet certain conditions on the server:
To enable installing over FTP, you need to allow FTP access to the installation directory on the server. This can be either anonymous FTP access or access through a named account with a password. If you want anonymous FTP to point to a different directory, you can use sym links to point to the installation directory on the server. The installation server is now ready to use. Make sure you note the server name or IP address and the directory path to the installation directory you created. Table of Contents If you experience odd behavior, application crashes, or have other issues, this chapter is meant to help you solve the problem if possible and, failing that, describes where the application logs are located and other information that can help your XenSource Solution Provider and XenSource track and resolve the issue. Troubleshooting of installation issues is covered in the XenServer Installation Guide. Troubleshooting of XenServer Host issues is covered in the XenServer Administrator's Guide. NoteWe recommend that you follow the troubleshooting information in this chapter solely under the guidance of your XenSource Solution Provider or XenSource Support. XenSource provides two forms of support: you can receive free self-help support via the Support site, or you may purchase our Support Services and directly submit requests by filing an online Support Case. Our free web-based resources include product documentation, a Knowledge Base, and discussion forums. If you are experiencing VM crashes, it's possible that a kernel crash dump might help identify the problem. If the crash is reproducible, follow this procedure to send the dumps to XenSource. For Linux VMs, the crashdump behavior can be controlled through the | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
