[Libguestfs] New tool proposal

[Richard W.M. Jones]

On Tue, May 22, 2012 at 08:01:50PM +0800, Wanlong Gao wrote:
> So, you means we can use virt-ls to list the different files list, and then
> use virt-cat to see the different between two single files?

Yes, this is what I usually do.

For example:

  $ qemu-img create -f qcow2 -b /dev/vg_pin/F16x64 cloned-f16.img
  Formatting 'cloned-f16.img', fmt=qcow2 size=21474836480 backing_file='/dev/vg_pin/F16x64' encryption=off cluster_size=65536 
  
  $ qemu-kvm -hda cloned-f16.img -m 1024
  [boot it, shut it down]
  
  $ virt-ls -lR -a /dev/vg_pin/F16x64 / > original.files     # *
  $ virt-ls -lR -a cloned-f16.img / > modified.files
  
  $ diff -u original.files modified.files
  --- original.files      2012-05-22 13:11:00.949456251 +0100
  +++ modified.files      2012-05-22 13:11:38.248133406 +0100
  @@ -1651,7 +1651,7 @@
   - 0644        373 /etc/rearj.cfg
   l 0777         14 /etc/redhat-release -> fedora-release
   - 0644       1787 /etc/request-key.conf
  -- 0644         79 /etc/resolv.conf
  +- 0644         74 /etc/resolv.conf
   l 0777         11 /etc/rmt -> ../sbin/rmt
   - 0644       1615 /etc/rpc
   d 0755       4096 /etc/rpm
  @@ -32090,6 +32090,7 @@
   d 0775       4096 /sys
   d 1777       4096 /tmp
   d 1777       4096 /tmp/.ICE-unix
  +s 0777          0 /tmp/.ICE-unix/1061
   s 0777          0 /tmp/.ICE-unix/1085
   d 1777       4096 /tmp/.Test-unix
   d 1777       4096 /tmp/.X11-unix
  @@ -207760,7 +207761,7 @@
   - 0644         28 /var/lib/AccountsService/users/rjones
   d 0755       4096 /var/lib/NetworkManager
   - 0644         87 /var/lib/NetworkManager/NetworkManager.state
  -- 0644       1454 /var/lib/NetworkManager/timestamps
  +- 0644       1502 /var/lib/NetworkManager/timestamps
   d 0755       4096 /var/lib/PackageKit
   - 0644       9216 /var/lib/PackageKit/transactions.db
   d 0755       4096 /var/lib/alsa
  
  [etc etc]

However there are some shortcomings.  It won't find differences in:

- filesystem metadata
- bootloaders or other information outside the regular filesystem
- inside the Windows Registry (it'll just say the two binary
  blobs are different)

Rich.

* You don't need the --checksum option unless files are modified but
  keep the same size, in which case you have to examine the complete
  content of each file.  Using --checksum is much slower.  'virt-ls'
  has other options that you may want to look at.

-- 
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/