When Kiwi builds an ISO, it is always supposed to be bootable and should
be located in the iso/ subdirectory.
Any other kind of image should still land in images/ and be listed as
not bootable in the metadata.
Relates: https://pagure.io/pungi-fedora/issue/1342
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
(cherry picked from commit d9d21d3cf4eaad5cc7f2959a4abdafed781bb9cf)
The version follows the same rules as versioning for live media etc.
That means it's always going to be set. The precedence goes like this:
* image specific option
* `kiwibuild_version`
* `global_version`
* `release_version` or `<release_version>_<label_milestone>`.
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
(cherry picked from commit d351773dab7b3aa8e6de82bbe23058b6b3448dd4)
PR #1789 improved various aspects of the ostree_container phase
regarding subvariant handling and filenames, this is mainly to
help us with how we want to handle bootc images (currently in
the IoT compose, but the generic base bootc image may move to
the Fedora compose).
PR #1790 rejigs the compose phase handling so the main image
build phase is not unnecessarily blocked on the ostree_install
phase. This should cut 60-90 minutes out of the main Fedora
compose time.
(cherry picked from commit 13884fef2c199b613442af03c24b737a7e3cb057)
We can have a compose with unsigned packages.
By the time the next compose is generated, the packages could have been
signed. However, the new compose would still reuse the ISO with unsigned
copies.
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
(cherry picked from commit d546a49299)
Without this fix existing configurations break even though they don't
use the phase.
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
(cherry picked from commit c59f2371a3)
Running xorriso to modify an ISO image can update content of included
images such as images/eltorito.img, unless we explicitly update the
image, which is undesirable (https://pagure.io/pungi/issue/1647).
However, when the file is changed, the checksum changes and .treeinfo no
longer matches.
This patch implements a workaround: once the DVD is written, it looks
for incorrect checksums, recalculates them and updates the .treeinfo on
the DVD. Since only the checksum is changing and the size of the file
remains the same, this seems to help fix the issue.
An additional step for implanting MD5 is needed again, as that gets
erased by the workaround.
JIRA: RHELCMP-13664
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
(cherry picked from commit 3b2c6ae72a)
This was only usable in live_images phase that doesn't exist anymore,
and wasn't used much in the first place.
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
(cherry picked from commit 0726a4dca7)
This phase was used to create live images with livecd-creator
and 32-bit ARM images with appliance-creator. We also remove
get_create_image_cmd from the Koji wrapper as it was only used
for this phase, remove associated tests, and remove related
configuration settings and documentation.
Fixes: https://pagure.io/pungi/issue/1753
Merges: https://pagure.io/pungi/pull-request/1774
Signed-off-by: Adam Williamson <awilliam@redhat.com>
(cherry picked from commit 531f0ef389)
* dict.sorted and funcsigs are not used anywhere anymore
* urlgrabber is used only in the yum based gather.py module, and thus
only needed on Python 2
* py3 doesn't need to reinstall mock as that is part of stdlib now
Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
(cherry picked from commit c96b5358ba)