* Mon Oct 27 2025 marcus.schaefer@suse.com
- Bump version: 2.1.25 → 2.1.26
* Mon Oct 27 2025 marcus.schaefer@suse.com
- Bump Migration changelog version: 2.1.26
* Thu Oct 23 2025 marcus.schaefer@suse.com
- Set systemd offline for Zypper in chroot mode
Calling zypper in a new system root should be done in
offline systemd mode to prevent package scripts to access
services not necessarily present
* Wed Oct 22 2025 marcus.schaefer@suse.com
- Fix apparmor install procedure
The apparmor to selinux unit install the selinux pattern after
the main migration has been performed. This zypper call has to
happen inside of the new root system otherwise repo access
is denied on e.g cloud systems which holds the registration
information in the installed system.
* Tue Oct 21 2025 marcus.schaefer@suse.com
- Fixed azuremetadata device lookup
On Azure the azuremetadata call is prepared with the --device
option pointing to the disk containing the system. This disk
device is obtained via the get_root_disk_device method which
makes use of the findmnt tool to read the device that contains
the root filesystem mountpoint. If that filesystem is using
sophisticated technology like subvolumes on btrfs, findmnt
reports this information in addition to the actual unix block
device in brackets [...]. For the logic flow of get_root_disk_device
this extra information broke the subsequent lsblk call. This
commit strips eventually existing extra block device information
from the findmnt call output.
* Mon Oct 20 2025 marcus.schaefer@suse.com
- Use of f-strings not allowed in the DMS
The code has to stay compatible with python 3.4 because
SLE12-to-SLE15 migrations has to continue to work. On
SLE12 only python 3.4 exists. This commit fixes the
code that used f-strings executed on python 3.4.
This is related to bsc#1248137
* Thu Oct 16 2025 jorik.cronenberg@suse.com
- Simplify interface naming by disabling predictable names at boot
Currently our most difficult problem with regards to the network interface
naming scheme is that when the migration image boots the interfaces get
predictable names like 'enp1s0' which we then have to revert/write custom rules
around. To simplify this we can install a file at
/etc/systemd/network/99-default.link. This skips all naming schemes and the
kernel names are used. Then we can remove this file again and copy over all the
necessary udev rules and link files from the host and with udevadm trigger apply
these. This should imitate the behavior of the final migrated system most
closely.
* Wed Oct 15 2025 marcus.schaefer@gmail.com
- Fixed test_check_lsm_migration unit test
The test did not mock shutil.which and thus was really
looking for the tool to be present on the system. If
not present on the system the test failed. This commit
fixes it.
* Wed Oct 15 2025 jorik.cronenberg@suse.com
- Fix setup_host_network_test by mocking os.makedirs
When '/etc/udev/' doesn't exist the test fails. But the call should be mocked
anyway and not actually executed on the system.
* Wed Oct 15 2025 cfamullaconrad@suse.com
- wicked2nm: log network state on nm-online failure
Version: 2.1.23-160000.1.1
* Fri Sep 19 2025 marcus.schaefer@suse.com
- Activation packages can no longer be noarch
As we have architectures that does not support the grub
based activation and therefore an ExclusiveArch setup in
the spec, the package can no longer be noarch
* Fri Sep 19 2025 marcus.schaefer@suse.com
- Bump version: 2.1.22 → 2.1.23
* Fri Sep 19 2025 marcus.schaefer@suse.com
- Bump Migration changelog version: 2.1.23
* Fri Sep 19 2025 marcus.schaefer@suse.com
- Fix update of image .changes files
We want a new entry with each version change but
bumpversion modifies the existing version changelog
entries which is unwanted. Thus the .changes files
got dropped from .bumpversion.cfg and a little
helper tool to handle the .changes entry was added
and called through the Makefile
* Fri Sep 19 2025 marcus.schaefer@suse.com
- Bump version: 2.1.21 → 2.1.22
* Fri Sep 19 2025 yu.daike@suse.com
- architecture check before allowing migration to be activated
SLE16 will panic on CPUs older than x86_64_v1 (i.e. x86_64). To
avoid panicing the system we check for the architecture before
activating the migration.
Co-authored-by: Marcus Schäfer <marcus.schaefer@suse.com>
* Thu Sep 18 2025 jorik.cronenberg@suse.com
- Improve error logs when wicked2nm fails
* Mon Sep 15 2025 marcus.schaefer@suse.com
- Bump version: 2.1.20 → 2.1.21
* Mon Sep 15 2025 marcus.schaefer@suse.com
- Include image changelog to version bump
Make sure the image changelog file also contains an entry
to match with the version of the DMS
* Mon Sep 15 2025 marcus.schaefer@suse.com
- Bump version: 2.1.19 → 2.1.20