Daniel P. Berrange wrote:
That posting is mostly based on a mis-understanding of what XenD is doing, but also the fact that there is no way to get the config back out of XenD using 'xm'. ie, you can load a traditional /etc/xen style config into XenD using 'xm new CONFIG', but you can't ask XenD to generate you a traditional style cofnig for one of the VMs its managing. Not an issue if you're using libvirt for management, since it doesn't use /etc/xen style configs anyway. While, there are a few people who need to use /etc/xen configs to get access to specific features not supported in libvirt (PCI device passthrough), they are a small niche. On the whole switching to using the new inactive domain management APIs in XenD was a net-win, with the benefits far outweighing the disadvantages.
Can we mix and match. Can we have some old style configs in "/etc/xen" for some guests while using libvirt for the rest?
I'm using the new way of doing things now but wonder what will happen if some day I need PCI device passthrough for one guest. Do, I have to revert all guests back to the old style of config to have this capability for one?
-- Fedora-xen mailing list Fedora-xen@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-xen
