Description of problem:
The version of libguestfs that we are shipping in RHEL 6.0 will be
1.2.7 + a handful of backported bug fixes and features to support
This bug is here to make the case that we should rebase libguestfs
in RHEL 6.1 to either:
>= 1.2.10 (along the same 1.2 stable branch), or:
>= 1.4.1 (along the new 1.4 stable branch)
Summary of libguestfs version numbers in all Fedora branches:
Explanation of libguestfs version numbering:
libguestfs 1.4.0 release notes:
Firstly, a rebase to >= 1.2.10 isn't really a "rebase" at all. We only
pull in bug fixes along the stable branches, so upgrading to this
version just pulls in bug fixes, some of which we are already carrying
as separate patches in the RHEL 6.0 SRPM. This is the completely
However a rebase to >= 1.4.1 should also be considered. This is
the latest stable branch. It has significant new features, but retains
source and binary (ABI) compatibility with existing programs.
Amongst the important new features for our users:
- virt-resize contains important fixes for Windows guests
which would be difficult to backport
- virt-resize contains useful extra capabilities for Linux guests
- the aug_clear API which we need in virt-v2v and currently
have to backport is included by default
- the FUSE module supports write mode
- guestfish has many usability enhancements
- a lot of more invasive bug fixes which were too much to
backport into 1.2 upstream are included in 1.4, and moving
to 1.4 directly would avoid us having to do these complex
backports just for RHEL.
Fedora 13 users are the test bed for the 1.4 branch.
The downsides as I see them:
- QA would have to be extended to cover the new features
- obviously there is some risk of regression with any big rebase
- virt-v2v is currently QA'd against 1.2
I am using Fedora 13 as a testbed for a possible RHEL 6.1 package.
This package is based on stable branch 1.4, with significant features
backported from the development branch. We are using git to
manage the patches.
Latest Fedora package, search for 'fc13' here:
Package spec file and patches:
The base version and backports are encoded in the version
f13$ guestfish --version
Amongst the significant enhancements over 1.2.7:
- big performance improvements
- use of virtio-serial instead of SLIRP
- all APIs required by virt-v2v are included or backported
- all bugfixes required by virt-v2v similarly
- virt-resize supports Windows guests
- FUSE module works in write mode
- other features are covered in the 1.4.0 release notes
We would like to rebase to 1.6. This includes all the
1.4 fixes, so the previous testing is still valid.
List of upstream feature and bug fix work is here:
There are some test packages here:
Setting back to ASSIGNED.
There are several problems with the packages from
comment 17 (however, *please* still test those):
1. Should package libguestfs-winsupport
2. Should require febootstrap 2.11
3. Need to backport the C tools from upstream for
RHEV-H and change the packaging slightly so that
nothing we need for RHEV-H will require perl.
(In reply to comment #19)
> 1. Should package libguestfs-winsupport
Turns out this is not required. You can continue to use the
current libguestfs-winsupport[a] package with this newer
> 2. Should require febootstrap 2.11
> 3. Need to backport the C tools from upstream for
> RHEV-H and change the packaging slightly so that
> nothing we need for RHEV-H will require perl.
Done. The new subpackage is called libguestfs-tools-c.
If you install libguestfs-tools (Perl tools) it will pull
in libguestfs-tools-c as well.
If you install libguestfs-tools-c, you will just get the
C tools and there should be no Perl dependency at all.
Further updates, see changelog:
An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on therefore solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.