For a very long time we have maintained two sets of utility functions,
in mllib/common_utils.ml and generator/utils.ml. This changes things
so that the same set of utility functions can be shared with both
directories.
It's not possible to use common_utils.ml directly in the generator
because it provides several functions that use modules outside the
OCaml stdlib. Therefore we add some lightweight post-processing which
extracts the functions using only the stdlib:
(*<stdlib>*)
...
(*</stdlib>*)
and creates generator/common_utils.ml and generator/common_utils.mli
from that. The effect is we only need to write utility functions
once.
As with other tools, we still have generator-specific utility
functions in generator/utils.ml.
Also in this change:
- Use String.uppercase_ascii and String.lowercase_ascii in place
of deprecated String.uppercase/String.lowercase.
- Implement String.capitalize_ascii to replace deprecated
String.capitalize.
- Move isspace, isdigit, isxdigit functions to Char module.
Support for RHEV with RHEL 6 nodes required us to output the old style
qcow2 compat=0.10 images. Since RHEV 3.6 GA, RHEL 6 has not been
supported as a RHEV node type. Since RHV 4.1, compat=1.1 is
supported. (Support for compat=1.1 is uncertain in RHV 4.0 even on
RHEL 7 nodes.)
There are significant downsides to using qcow2 compat=0.10 instead of
the modern default (compat=1.1).
Therefore this patch does two things:
For -o rhev, it drops support for compat=0.10 completely. You must
use RHV 4.1.
For -o vdsm, it adds an interim flag (--vdsm-compat=0.10 or
--vdsm-compat=1.1) which controls the compat level of the qcow2 output
file. VDSM should use --vdsm-compat=1.1 when it is known that modern
qemu is available. We can make this the default later when all RHV
instances have moved to 4.1.
It also adds:
vdsm-compat-option
to the `virt-v2v --machine-readable' output to indicate that this flag
can be used.
Thanks: Yaniv Kaul, Michal Skrivanek.
Flag the filesystems for Linux /usr properly as USR role, and detect
some data out of it, like the distro information from an os-release
(if present), and the architecture (since the binaries used for our
architecture check will be available there only).
Later on, collect the results in a way similar to what is done for
CoreOS: for each non-CoreOS root, try to find its /usr filesystem, and
if found then merge what is missing from root; in the last case, also
override the distro inspection data (version, product name) if available
in /usr.
When checking for the existance of /dev/mapper devices found in the
fstab of a filesystem, using guestfs_exists means they are checked as
files in the guest, while they really appear as devices on the
appliance. Instead, try the lvm name resolution anyway, and ignore them
when they are reported as missing.
Thanks to: Richard W.M. Jones.
Fixes commit 96b6504b09.
Move the checks for empty xmlXPathObjectPtr, and for extracting the
result string out of it, to a new helper functions.
This is just code motion, there should be no behaviour changes.
Add a new inspect role for "/usr" partitions, and use that to mark the
/usr partition in CoreOS: this additional role allows to ease its lookup
later on, when merging its results into those of the root.
Introduce a new enum to classify the role of a filesystem, if available.
This will help later on when doing operations on non-root filesystems,
like detecting particular mountpoints such as /usr.
The new enum has only "root" as known role, which replaces the is_root
flag.
Tools could require the use of pseudo-terminals, so make sure we have
/dev/pts available in the appliance. The "command" API already
bind-mounts it when running commands, so this is the only bit needed.
This refactors command line parsing into a parse_cmdline function.
The function uses the Arg module to parse the command line instead of
ad hoc parsing.
This also enforces use of the ./run script to run the program, since
otherwise you can end up with mixed versions of the OCaml bindings and
the C library.
This updates commit 65a0570385.
initviocons package provides tools to resize the terminal. Having it
in the appliance will allow SUSE users to have proper line wrapping
in their terminal when using virt-rescue.
Rebuild these images with the new make-template.ml script. This fixes
the lack of virtio-scsi driver and broken initramfs.
These images all use LVM because without that both Fedora and RHEL put
the root filesystem on an extended partition (/dev/sda5) which
virt-resize cannot handle.
(Note the RHEL images are not public.)
Create a new directory (builder/template). Integrate all of the
scripts into a single program, so that templates are generated more
consistently.
This also changes how the index file is generated. The script now
generates the index file fragment and saves it under version control,
and then generates the final index file by concatenating these.
(Previously the index was written by hand which was tedious and
error-prone.)
The new script also saves the generated kickstart under version
control so it can be referenced later.
It turns out that adding the virtio-scsi driver does not help
to fix the dracut problem.
This partially reverts commit f766c84c39,
keeping the change to index.
Unfortunately I was unable to build s390 binaries since multilib was
dropped in Fedora 24 on s390x. Going from the source of the 'file'
command it seems as if it prints "32-bit" (the architecture is really
31 bit).
As noted by Pino in another patch, the logic passes the first member of
the struct which happens to be the right address to the callback
function.
This will break the callback if order of the members of the struct will
change.
As the callback is using the entire struct, better to pass the pointer
to the struct itself.
Signed-off-by: Matteo Cafasso <noxdafox@gmail.com>
If an environment variable such as XDG_RUNTIME_DIR or one of the
tmpdirs or cachedir is set to a non-existent directory, improve the
error message that the user will see so that (where possible) it
includes the environment variable or API call.
This is still not bullet-proof because it's hard to display the
environment variable if it is LIBGUESTFS_TMPDIR or
LIBGUESTFS_CACHEDIR, but the main problem is with XDG_RUNTIME_DIR
(because of systemd bugs).
Thanks: Hilko Bengen for identifying the bug.
