11:09 a.m.
Hi all.
I've been playing around with KVM for a bit now and have several virtual
machines up and running. Having passed what I considered "first look"
stage, I'm ready to start using libvirt as a manager. For reference,
pre-libvirt, I have started this VM with:
/kvm -net nic,model=virtio -net tap,ifname=tap5,script=no -drive
file=test1.img,if=virtio,boot=on -curses -no-reboot -m 1g -smp 2/
I have a test environment I like to play with, and created the initial
domain XML using virt-install:
/virt-install --name "Test" --ram 1024 --vcpus=2 --cpu host
--description "Test VM" --import --file test1.img --os-type=linux
--os-variant=virtio26 --network bridge=br0,model=virtio --graphics=none
--autostart --prompt/
When I look in /var/log/libvirt/qemu/Test.log I see the VM is started
with the command (edited down to the relevant parts):
//usr/bin/qemu-kvm -S -m 1024 -smp 2 -name Test,process=qemu:Test
-nographics -boot c -drive file=/kvms/test1.img,if=none,boot=on
/This has two problems:
1) The "-S" is keeping the system from booting
2) the "if=none" needs to be "if=virtio"
I've confirmed both by changing -nographics to -curses and watching what
happens. For #1, its a big fat nothing (no surprise since -S turns off
the CPUs), for #2 grub loads and the kernel boots, but my /dev/vda3
isn't found so "init" can't start.
I'm anticipating a 3rd problem: I don't see where to specify that I
want the VM to use "tap5" of my bridged network.
Can some kind soul please let me know what I need to change in the
domain XML file to fix these issues?
Thanks,
Kevin
ps. I've post this to my blog at
http://www.kevinsthoughts.com/2011/08/kvm-management-libvirt/ if anyone
would prefer to answer there.
11:14 p.m.
On 08/29/2011 08:09 PM, Kevins Thoughts wrote:
This is intentional. Libvirt has to make some changes via the monitor,
since qemu does not expose all functionality from the command line, so
libvirt blindly starts qemu with -S, makes subsequent monitor changes,
and then released the vcpus via monitor command (unless you used the
VIR_DOMAIN_START_PAUSED flag to keep things paused). There is no XML to
control this, nor does there need to be, given the existence of
virDomainCreateWithFlags().
2) the "if=none" needs to be "if=virtio"
Have you tried 'virsh domxml-from-native qemu-argv /path/to/args' for a
shell-style command line containing the qemu command you want to run?
It's not perfect, but may give you some hints on this regard.
Meanwhile, according to the documentation, <target dev=...
bus='virtio'/> inside your <drive> xml appears to be the right way to
request a virtio drive. Paste the actual XML you are trying to use, so
we can offer suggestions relative to your questions.
Does the documentation help?
http://libvirt.org/formatdomain.html#elementsNICS
--
Eric Blake eblake(a)redhat.com +1-801-349-2682
Libvirt virtualization library http://libvirt.org
12:59 a.m.
On 8/30/2011 9:14 AM, Eric Blake wrote:
Alas, for whatever reason, the vcpus do not appear to be being
released. XML below...
No I haven't, thanks for that suggestion.
Actually, I did read the documentation and agree - at least that was my
best guess, but its not working. The generated command being invoked
has "-drive file=/kvms/test1.img,if=none,boot=on" specified.
Here is the XML file created by the command: virt-install --name "Test"
--ram 1024 --vcpus=2 --cpu host --description "Test VM" --import --file
test1.img --os-type=linux --os-variant=virtio26 --network
bridge=br0,model=virtio --graphics=none --autostart --prompt
<Test.xml>
<domain type='kvm'>
<name>Test</name>
<uuid>b0784171-0662-80ce-caad-c9e3c7fe0952</uuid>
<description>Test VM</description>
<memory>1048576</memory>
<currentMemory>1048576</currentMemory>
<vcpu>2</vcpu>
<os>
<type arch='x86_64' machine='pc-0.13'>hvm</type>
<boot dev='hd'/>
</os>
<features>
<acpi/>
<apic/>
<pae/>
</features>
<cpu match='exact'>
<model>Opteron_G3</model>
<vendor>AMD</vendor>
<feature policy='require' name='skinit'/>
<feature policy='require' name='vme'/>
<feature policy='require' name='mmxext'/>
<feature policy='require' name='fxsr_opt'/>
<feature policy='require' name='cr8legacy'/>
<feature policy='require' name='ht'/>
<feature policy='require' name='3dnowprefetch'/>
<feature policy='require' name='3dnowext'/>
<feature policy='require' name='wdt'/>
<feature policy='require' name='extapic'/>
<feature policy='require' name='pdpe1gb'/>
<feature policy='require' name='osvw'/>
<feature policy='require' name='cmp_legacy'/>
<feature policy='require' name='3dnow'/>
</cpu>
<clock offset='utc'/>
<on_poweroff>destroy</on_poweroff>
<on_reboot>restart</on_reboot>
<on_crash>restart</on_crash>
<devices>
<emulator>/usr/bin/qemu-kvm</emulator>
<disk type='file' device='disk'>
<driver name='qemu' type='raw'/>
<source file='/kvms/test1.img'/>
<target dev='vda' bus='virtio'/>
<address type='pci' domain='0x0000' bus='0x00'
slot='0x04' function='0x0'/>
</disk>
<interface type='bridge'>
<mac address='52:54:00:f3:28:7b'/>
<source bridge='br0'/>
<model type='virtio'/>
<address type='pci' domain='0x0000' bus='0x00'
slot='0x03' function='0x0'/>
</interface>
<serial type='pty'>
<target port='0'/>
</serial>
<console type='pty'>
<target type='serial' port='0'/>
</console>
<memballoon model='virtio'>
<address type='pci' domain='0x0000' bus='0x00'
slot='0x05' function='0x0'/>
</memballoon>
</devices>
</domain>
Does the documentation help?
http://libvirt.org/formatdomain.html#elementsNICS
Looks like I could perhaps add a "<target dev='tap5'/>" to the
bridge
interface section. The documentation refers to using "tun" devices, and
I use "tap" devices, but perhaps that's close enough.
4965
days inactive
4965
days old
2 comments
2 participants
participants (2)
-
Eric Blake -
Kevins Thoughts