KVM and Libvirt - How do I hotplug a new virtio disk?

Question

I am trying to hot-add a file-based disk to a running KVM virtual server. I've created a new disk from scratch using the command

dd of=/home/cloud/vps_59/test.img bs=1 seek=5G count=0

and I was hoping to get it hot-added to the guest by doing this in the virsh shell:

virsh # attach-disk vps_59 /home/cloud/vps_59/test.img \
        vdd --driver=file --subdriver=raw

The XML definition of the domain then becomes:

<disk type='file' device='disk'>
  <driver name='qemu' type='raw'/>
  <source file='/home/cloud/vps_59/root.img'/>
  <target dev='vda' bus='virtio'/>
</disk>
<disk type='file' device='disk'>
  <driver name='file' type='raw'/>
  <source file='/home/cloud/vps_59/test.img'/>
  <target dev='vdd' bus='virtio'/>
</disk>

As you can see, the driver name becomes wrong, it should be driver name='qemu' as the existing vda disk. I have tried with --drive=qemu but it states it is unsupported.

Secondly, I only "see" the newly added drive once I reboot the virtual machine running Ubuntu 10.04.4 LTS. How can I make the drive "hotplug"? I want the virtual machine to "see" the new drive immediately without a reboot.

Answer 1

I'd like to start with a note that you should avoid using virsh attach-disk with its limited amount of options. Instead, I'd suggest to specify the exact disk format you prefer in a separate, temporary XML file or by using the virt-manager GUI application (for the latter, skip the first step).

1. Create a temporary file with a disk definition like this one below.

   adjust the properties to fit your situation

   <disk type='file' device='disk'>
     <driver name='qemu' type='qcow2'/>
     <source file='/path/to/disk-image.img'/>
     <target dev='vdb' bus='virtio'/>
   </disk>
   

   Tip: Peek into your current XML domain configuration and copy a <disk> section from there.

   virsh dumpxml <domainname>
   

2. Now, before adding the disk to a current domain, make sure the required hotplug kernel modules are loaded in the guest.

   Some Linux distributions like recent CentOS/RHEL/Fedora have this built-in in the kernel. In this case, check for CONFIG_HOTPLUG_PCI_ACPI. If it's y, then you're all set and you can skip this step.

   modprobe acpiphp
   modprobe pci_hotplug
   

   Consider adding these two modules to /etc/modules if you want them to be loaded on boot by default.

3. Add the disk it to the running VM using

   virsh attach-device <domain name> /path/to/disk.xml
   

   Optionally, add the --persistent option to let Libvirt update the domain XML definition 'persistent'.

4. Finally, check inside the guest if the disk was indeed hotplug-inserted. The kernel should be triggered, as can be checked with dmesg:

   [  321.946440] virtio-pci 0000:00:06.0: using default PCI settings
   [...]
   [  321.952782]  vdb: vdb1 vdb2
   

   In the above example I've added a disk as vdb with two partitions in the partition table.

References

• Linux-KVM wiki - Using PCI Hotplug Support
• similar question "Adding Virtio block devices at runtime in Libvirt KVM"