Re: [Libguestfs] guestfish/libguestfs takes legacy qemu instead of kvm?
On Sun, Aug 08, 2010 at 12:05:58PM +0800, Kirby Zhou wrote:
guestfish takes legacy qemu instead of kvm? ]# rpm -q libguestfs --requires | fgrep qemu qemu-system-x86 >= 0.10.5
To make it clear, this is EPEL-5. There was some still unresolved problem with adding a dependency on KVM: https://fedorahosted.org/rel-eng/ticket/2982 As a result, we have to depend on qemu (software emulation) instead. However you can easily change this at runtime by setting the LIBGUESTFS_QEMU environment variable: http://libguestfs.org/guestfs.3.html#libguestfs_qemu http://libguestfs.org/guestfs.3.html#qemu_wrappers In RHEL 6 it's all going to be different again. The same qemu-system-x86 RPM is actually KVM. We combine qemu and KVM sources during the build to make a 'qemu-kvm' binary which can do both software emulation (TCG) and hardware virt (KVM).
[@djt_10_48 ~]# guestfish -i kvm-rhel5guest Could not open '/dev/kqemu' - QEMU acceleration layer not activated: No such file or directory
This is just a warning -- ignore it. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones virt-df lists disk usage of guests without needing to install any software inside the virtual machine. Supports Linux and Windows. http://et.redhat.com/~rjones/virt-df/
Thanks very much. BTW, I have a small question, why libguestfs depends on QEMU instead of libvirt? If we take libvirt, we can easily run libguestfs with xen, virtualbox, vmware, etc. Regards, Kirby Zhou from SOHU-RD +86-10-6272-8261 -----Original Message----- From: libguestfs-bounces@redhat.com [mailto:libguestfs-bounces@redhat.com] On Behalf Of Richard W.M. Jones Sent: Sunday, August 08, 2010 7:26 PM To: EPEL development disccusion Cc: libguestfs@redhat.com Subject: Re: [Libguestfs] guestfish/libguestfs takes legacy qemu instead of kvm? On Sun, Aug 08, 2010 at 12:05:58PM +0800, Kirby Zhou wrote:
guestfish takes legacy qemu instead of kvm? ]# rpm -q libguestfs --requires | fgrep qemu qemu-system-x86 >= 0.10.5
To make it clear, this is EPEL-5. There was some still unresolved problem with adding a dependency on KVM: https://fedorahosted.org/rel-eng/ticket/2982 As a result, we have to depend on qemu (software emulation) instead. However you can easily change this at runtime by setting the LIBGUESTFS_QEMU environment variable: http://libguestfs.org/guestfs.3.html#libguestfs_qemu http://libguestfs.org/guestfs.3.html#qemu_wrappers In RHEL 6 it's all going to be different again. The same qemu-system-x86 RPM is actually KVM. We combine qemu and KVM sources during the build to make a 'qemu-kvm' binary which can do both software emulation (TCG) and hardware virt (KVM).
[@djt_10_48 ~]# guestfish -i kvm-rhel5guest Could not open '/dev/kqemu' - QEMU acceleration layer not activated: No such file or directory
This is just a warning -- ignore it. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones virt-df lists disk usage of guests without needing to install any software inside the virtual machine. Supports Linux and Windows. http://et.redhat.com/~rjones/virt-df/ _______________________________________________ Libguestfs mailing list Libguestfs@redhat.com https://www.redhat.com/mailman/listinfo/libguestfs
On 10/08/10 10:27, Kirby Zhou wrote:
Thanks very much. BTW, I have a small question, why libguestfs depends on QEMU instead of libvirt? If we take libvirt, we can easily run libguestfs with xen, virtualbox, vmware, etc.
Not quite. There's more to making an appliance run on a target hypervisor than just creating an appropriate domain. The appliance itself must support the target. However, there's an argument to be made about using libvirt to isolate ourselves from the whims of qemu command-line argument changes. Matt -- Matthew Booth, RHCA, RHCSS Red Hat Engineering, Virtualisation Team GPG ID: D33C3490 GPG FPR: 3733 612D 2D05 5458 8A8A 1600 3441 EA19 D33C 3490
On Tue, Aug 10, 2010 at 05:27:56PM +0800, Kirby Zhou wrote:
Thanks very much. BTW, I have a small question, why libguestfs depends on QEMU instead of libvirt? If we take libvirt, we can easily run libguestfs with xen, virtualbox, vmware, etc.
It's considerably more complex than that. However you can already use libguestfs to analyze images from xen, vmware, virtualbox and so on. The fact that qemu is involved behind the scenes doesn't matter -- indeed it's an advantage because qemu's block layer supports many different container formats. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones virt-p2v converts physical machines to virtual machines. Boot with a live CD or over the network (PXE) and turn machines into Xen guests. http://et.redhat.com/~rjones/virt-p2v
But pure QEMU is very very slow. Regards Kirby Zhou -----Original Message----- From: Richard W.M. Jones [mailto:rjones@redhat.com] Sent: Tuesday, August 10, 2010 8:34 PM To: Kirby Zhou Cc: 'EPEL development disccusion'; libguestfs@redhat.com Subject: Re: [Libguestfs] guestfish/libguestfs takes legacy qemu instead of kvm? On Tue, Aug 10, 2010 at 05:27:56PM +0800, Kirby Zhou wrote:
Thanks very much. BTW, I have a small question, why libguestfs depends on QEMU instead of libvirt? If we take libvirt, we can easily run libguestfs with xen, virtualbox, vmware, etc.
It's considerably more complex than that. However you can already use libguestfs to analyze images from xen, vmware, virtualbox and so on. The fact that qemu is involved behind the scenes doesn't matter -- indeed it's an advantage because qemu's block layer supports many different container formats. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones virt-p2v converts physical machines to virtual machines. Boot with a live CD or over the network (PXE) and turn machines into Xen guests. http://et.redhat.com/~rjones/virt-p2v
On Wed, Aug 11, 2010 at 12:10:13AM +0800, Kirby Zhou wrote:
From: Richard W.M. Jones [mailto:rjones@redhat.com]
On Tue, Aug 10, 2010 at 05:27:56PM +0800, Kirby Zhou wrote:
Thanks very much. BTW, I have a small question, why libguestfs depends on QEMU instead of libvirt? If we take libvirt, we can easily run libguestfs with xen, virtualbox, vmware, etc.
It's considerably more complex than that.
But pure QEMU is very very slow.
OK, so I think what you're asking for is to be able to run libguestfs with KVM (ie. hardware acceleration) either alongside another hypervisor (eg. Xen) or inside another virtual machine. This [hardware acceleration] is indeed not possible, and so you will fall back to using software emulation (ie. pure TCG-based qemu), and this is somewhat slow, although at least it works! The only solution to this is some sort of nested virt. Since I am at the KVM Forum at the moment, I have been emphasizing the importance to us of having nested virt at least in KVM. But the implementation of nested virt is very complex. Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming blog: http://rwmj.wordpress.com Fedora now supports 80 OCaml packages (the OPEN alternative to F#) http://cocan.org/getting_started_with_ocaml_on_red_hat_and_fedora
participants (3)
-
Kirby Zhou -
Matthew Booth -
Richard W.M. Jones