Migration Guide To Xen4.1+: Difference between revisions

From Xen
Jump to navigationJump to search
mNo edit summary
(fix link formatting)
Line 4: Line 4:
<!-- Page revision: 00000016 -->
<!-- Page revision: 00000016 -->
<!-- Original date: Sun Mar 27 08:26:16 2011 (1301214376000000) -->
<!-- Original date: Sun Mar 27 08:26:16 2011 (1301214376000000) -->

{{Needs_Formatting|Has some Wiki conversion formatting issues that need fixing, e.g. <pre>[[self:HostConfiguration/Networking|HostConfiguration]/Networking]</pre>}}


= Migration Guide To Xen 4.1+ =
= Migration Guide To Xen 4.1+ =
Line 30: Line 28:
* Managed domains (e.g. <code><nowiki>xm vm-create</nowiki></code>, <code><nowiki>xm vm-start</nowiki></code> etc) are not supported by <code><nowiki>xl</nowiki></code>. Other toolstacks which support managed domains are [http://www.xen.org/products/cloudxen.html Xen Cloud Platform's] <code><nowiki>xapi</nowiki></code> toolstack and [http://libvirt.org/ libvirt].
* Managed domains (e.g. <code><nowiki>xm vm-create</nowiki></code>, <code><nowiki>xm vm-start</nowiki></code> etc) are not supported by <code><nowiki>xl</nowiki></code>. Other toolstacks which support managed domains are [http://www.xen.org/products/cloudxen.html Xen Cloud Platform's] <code><nowiki>xapi</nowiki></code> toolstack and [http://libvirt.org/ libvirt].


The other notable difference is that <code><nowiki>xl</nowiki></code>, unlike <code><nowiki>xend</nowiki></code>, will not perform any host networking configuration for you. The <code><nowiki>xend</nowiki></code> code to reconfigure on startup is extremely fragile, prone to bugs when either distro or Xen are upgraded and rather difficult to debug. Instead <code><nowiki>xl</nowiki></code> takes the approach of deferring such network configuration to the distribution configuration tools which are usually well understood and documented. Therefore upon upgrading to <code><nowiki>xl</nowiki></code> it is necessary to configure your host networking stack appropriately. Please see [[self:HostConfiguration/Networking|HostConfiguration]/Networking] for more details on how to do this.
The other notable difference is that <code><nowiki>xl</nowiki></code>, unlike <code><nowiki>xend</nowiki></code>, will not perform any host networking configuration for you. The <code><nowiki>xend</nowiki></code> code to reconfigure on startup is extremely fragile, prone to bugs when either distro or Xen are upgraded and rather difficult to debug. Instead <code><nowiki>xl</nowiki></code> takes the approach of deferring such network configuration to the distribution configuration tools which are usually well understood and documented. Therefore upon upgrading to <code><nowiki>xl</nowiki></code> it is necessary to configure your host networking stack appropriately. Please see [[HostConfiguration/Networking]] for more details on how to do this.


Please report any other configuration file incompatibilities to the Xen developers on the [mailto:xen-devel@lists.xensource.com xen-devel mailing list].
Please report any other configuration file incompatibilities to the Xen developers on the [mailto:xen-devel@lists.xensource.com xen-devel mailing list].
Line 36: Line 34:
==== xl Upgrade Checklist ====
==== xl Upgrade Checklist ====


* Configure your host networking using the configuration tools provided by your distribution. (See: [[self:HostConfiguration/Networking|HostConfiguration]/Networking])
* Configure your host networking using the configuration tools provided by your distribution. (See: [[HostConfiguration/Networking]])
* Remove any python code from domU configuration files.
* Remove any python code from domU configuration files.
* Disable xend startup script
* Disable xend startup script
Line 45: Line 43:
The upgrade form previous <code><nowiki>xend</nowiki></code> versions is intended to be transparent, with the following exception:
The upgrade form previous <code><nowiki>xend</nowiki></code> versions is intended to be transparent, with the following exception:


* In previous releases <code><nowiki>xend</nowiki></code> unconditionally reconfigures the host network stack on startup, according to the <code><nowiki>network-script</nowiki></code> configuration item in <code><nowiki>/etc/xen/xend-config.sxp</nowiki></code>. In this release <code><nowiki>xend</nowiki></code> will only reconfigure the host network stack if the network stack does not appear to have been configured already (e.g. no bridges currently exist). This change allows administrators who wish to configure the network stack themselves (e.g. following the guidance on [[self:HostConfiguration/Networking|HostConfiguration]/Networking]) to do so by default while preserving the existing behaviour for those who do not. In addition this new behaviour allows for smoother transitions back and forth between <code><nowiki>xend</nowiki></code> and <code><nowiki>xl</nowiki></code> (as described in the <code><nowiki>xl</nowiki></code> upgrade notes) since it means you can safely make the necessary configuration changes to use <code><nowiki>xl</nowiki></code> and not have to undo them if you switch back to <code><nowiki>xend</nowiki></code>. In this case it is recommended to use the <code><nowiki>xenbrN</nowiki></code> naming convention for maximum compatibility.
* In previous releases <code><nowiki>xend</nowiki></code> unconditionally reconfigures the host network stack on startup, according to the <code><nowiki>network-script</nowiki></code> configuration item in <code><nowiki>/etc/xen/xend-config.sxp</nowiki></code>. In this release <code><nowiki>xend</nowiki></code> will only reconfigure the host network stack if the network stack does not appear to have been configured already (e.g. no bridges currently exist). This change allows administrators who wish to configure the network stack themselves (e.g. following the guidance on [[HostConfiguration/Networking]]) to do so by default while preserving the existing behaviour for those who do not. In addition this new behaviour allows for smoother transitions back and forth between <code><nowiki>xend</nowiki></code> and <code><nowiki>xl</nowiki></code> (as described in the <code><nowiki>xl</nowiki></code> upgrade notes) since it means you can safely make the necessary configuration changes to use <code><nowiki>xl</nowiki></code> and not have to undo them if you switch back to <code><nowiki>xend</nowiki></code>. In this case it is recommended to use the <code><nowiki>xenbrN</nowiki></code> naming convention for maximum compatibility.


[[Category:Xen]]
[[Category:Xen]]

Revision as of 01:33, 24 November 2011


Migration Guide To Xen 4.1+

Xen 4.1.0 includes a new toolstack xl. This new toolstack is intended to be a drop in replacement for the xm command line interface to the xend toolstack. xl is built on the libxl toolstack library which is intended as a common library which will be used by all toolstacks in the future.

The xend toolstack remains supported in Xen 4.1.0 however users are strongly recommended to upgrade to xl.

General upgrade notes

In this release the Xen build system will no longer clone and build a kernel on behalf of the user. There is no coupling betweeen a Xen release and the kernel used in either dom0 or domU nor any coupling between the dom0 kernel and the domU kernel.

For guest domain (domU) use users are strongly encouraged to take advantage of the Xen guest support which is present in many mainstream distributions. See XenDomUSupport for more information.

For domain 0 use please see the XenDom0Kernels page for more information and guidance.

Toolstack upgrade notes

xl upgrade notes and migration steps from xend

The xl toolstack is intended to remain backwards compatible with existing xm domain configuration files. It is therefore expected that any existing configurations should work with xl with the following exceptions:

  • Arbitrary embedded python is not supported. Therefore it will be necessary to remove any python code from your configuration files.
  • Managed domains (e.g. xm vm-create, xm vm-start etc) are not supported by xl. Other toolstacks which support managed domains are Xen Cloud Platform's xapi toolstack and libvirt.

The other notable difference is that xl, unlike xend, will not perform any host networking configuration for you. The xend code to reconfigure on startup is extremely fragile, prone to bugs when either distro or Xen are upgraded and rather difficult to debug. Instead xl takes the approach of deferring such network configuration to the distribution configuration tools which are usually well understood and documented. Therefore upon upgrading to xl it is necessary to configure your host networking stack appropriately. Please see HostConfiguration/Networking for more details on how to do this.

Please report any other configuration file incompatibilities to the Xen developers on the xen-devel mailing list.

xl Upgrade Checklist

  • Configure your host networking using the configuration tools provided by your distribution. (See: HostConfiguration/Networking)
  • Remove any python code from domU configuration files.
  • Disable xend startup script
  • Use xl command instead of xm

xend upgrade notes

The upgrade form previous xend versions is intended to be transparent, with the following exception:

  • In previous releases xend unconditionally reconfigures the host network stack on startup, according to the network-script configuration item in /etc/xen/xend-config.sxp. In this release xend will only reconfigure the host network stack if the network stack does not appear to have been configured already (e.g. no bridges currently exist). This change allows administrators who wish to configure the network stack themselves (e.g. following the guidance on HostConfiguration/Networking) to do so by default while preserving the existing behaviour for those who do not. In addition this new behaviour allows for smoother transitions back and forth between xend and xl (as described in the xl upgrade notes) since it means you can safely make the necessary configuration changes to use xl and not have to undo them if you switch back to xend. In this case it is recommended to use the xenbrN naming convention for maximum compatibility.