QEMU Storage Daemon

qemu-storage-daemon is a tool that provides disk image functionality for a VM without running the VM itself. It can export disk images, run block job operations, and perform other disk-related operations. This article was very helpful to get a better understanding of the QEMU storage daemon.

Here, I used two of the export methods for qsd to access the Fedora disk image that was created in the previous blog post.

1. Fuse export

FUSE (Filesystem in Userspace) is an interface for userspace programs to export a filesystem to the Linux kernel.

There is a really nice tutorial on FUSE export on the QEMU website which I followed through.

1. First of all, install some dependencies:
   

   sudo apt install libfuse3-3 libfuse3-dev
   

2. Then configure qemu with --enable-fuse option and build it again.
   

   ./configure --enable-fuse
   

3. Also, uncomment user_allow_other in /etc/fuse.conf. By default, only user who exported the image will have an access to it. Enabling user_allow_other option here will allow non-root users to be able to use allow_other mount option which enables users other than the owner to be able to access the mounted fuse image.

4. Create a file that will be a mount point where the filesystem will be mounted on. It's also possible to mount the image on itself.
   

   touch mount-point
   

5. Export:
   

   qemu-storage-daemon \
      --blockdev node-name=prot-node,driver=file, 
   filename=fedora37.qcow2   \
      --blockdev node-name=fmt-node,driver=qcow2,file=prot-node   
      --export type=fuse,id=exp0,node-name=fmt-node, 
   mountpoint=mount-point,writable=on  
   

   After the above command is executed, the mount-point will appear as a raw VM disk. Then, we'll be able to use tools like parted, kpartx, file,mount as they can't parse qcow2 images and can only work with raw images.
   

   $ qemu-img info mount-point
   image: mount-point
   file format: raw
   virtual size: 30 GiB (32212254720 bytes)
   disk size: 1.58 GiB
   

6. List the partition table:
   

   $ parted mount-point p
   WARNING: You are not superuser.  Watch out for permissions.
   Model:  (file)
   Disk /home/amar/projects/outreachy/qemu/fedora/mount-point: 32.2GB
   Sector size (logical/physical): 512B/512B
   Partition Table: gpt
   Disk Flags: 
   
   Number  Start   End     Size    File system  Name     Flags
    1      1049kB  2097kB  1049kB               primary  bios_grub
    2      2097kB  1076MB  1074MB  xfs          primary
    3      1076MB  32.2GB  31.1GB  xfs          primary
   

7. Add device mappings over partitions:
   

   $ sudo kpartx -av mount-point 
   [sudo] password for amar: 
   add map loop31p1 (253:6): 0 2048 linear 7:31 2048
   add map loop31p2 (253:7): 0 2097152 linear 7:31 4096
   add map loop31p3 (253:8): 0 60810752 linear 7:31 2101248
   

8. Mount the main partition:
   

   $ sudo mount /dev/mapper/loop31p3 /mnt/tmp/
   $ ls /mnt/tmp/
   afs  boot  etc   lib    media  opt   root  sbin  sys  usr
   bin  dev   home  lib64  mnt    proc  run   srv   tmp  var
   

9. Unmount the partition after done using it.
   

   sudo umount /mnt/tmp
   

10. Stop qsd by just hitting Ctrl+C or if it's running in the background, just kill the process.

2. NBD export

NBD (Network Block Device) is a simple protocol that allows users to export a block device to a client over a network. Here is a nice introduction of NBD as suggested by my mentors.

1. First, load NBD module:
   

   sudo modprobe nbd
   

2. Export the image file over NBD. With the below settings, NBD server will start running on localhost port 10809.
   

   qemu-storage-daemon     \
       --blockdev node-name=prot-node,driver=file,
   filename=fedora37.qcow2     \
       --blockdev node-name=fmt-node,driver=qcow2,
   file=prot-node     \
       --nbd-server addr.type=inet,addr.host=localhost,
   addr.port=10809     \
       --export type=nbd,id=exp0,node-name=fmt-node,
   name=guest-disk,writable=on
   

3. Mount the NBD exported device as a local device /dev/nbd0 by starting the nbd-client. For -N or -name option, use the name provided in the export command above. If this option is not supplied, nbd-client will complain about oldstyle negotiation.
   

   $ sudo nbd-client localhost 10809 /dev/nbd0 -N guest-disk
   [sudo] password for amar: 
   Negotiation: ..size = 30720MB
   Connected /dev/nbd0
   

4. Block devices for each partition of the original image will be automatically created.
   

   $ ls /dev/nbd0<tab>
   nbd0    nbd0p1  nbd0p2  nbd0p3
   

5. Mount the main partition
   

   $ sudo mount /dev/nbd0p3 /mnt/tmp/
   $ ls /mnt/tmp/
   afs  boot  etc   lib    media  opt   root  sbin  sys  usr
   bin  dev   home  lib64  mnt    proc  run   srv   tmp  var
   

6. After finished, unmount and disconnect. Proper clean-up is important because if not done correctly, the next export will not work.
   

   umount /mnt/tmp
   nbd-client -d /dev/nbd0