If the original ISO contains both `efiboot.img` and `macboot.img`,
the current `xorrisofs` invocation does not configure `efiboot.img`
to be bootable, only `macboot.img`.
This fix uses `>=` to compare `efimode` against `EFIBOOT`, since
`MACBOOT` implies `EFIBOOT` as well.
This is similar to the logic for `mkisofs`.
Fixes#1046.
Generated image confirmed to work, when comparing the reports:
`xorriso -indev $PATH_TO_ISO -report_el_torito cmd`
```
❯ diff -u ~/xorriso-report-{orig,new}.log
--- /home/michel/xorriso-report-orig.log 2020-07-15 20:01:01.405289018 -0700
+++ /home/michel/xorriso-report-new.log 2020-07-15 20:01:19.183555423 -0700
@@ -1,6 +1,6 @@
--volid 'Fedora-S-dvd-x86_64-32'
--volume_date uuid '2020042219432600'
--boot_image isolinux system_area=--interval:imported_iso:0s-15s:zero_mbrpt,zero_gpt:'/home/michel/Downloads/Fedora-Server-netinst-x86_64-32-1.6.iso'
+-volid 'Fedora-32-btrfs'
+-volume_date uuid '2020071602410900'
+-boot_image isolinux system_area=--interval:imported_iso:0s-15s:zero_mbrpt,zero_gpt:'/home/michel/src/github/michel-slm/luks-kickstarts/iso/Fedora-32-WS-20200715-3.iso'
-boot_image any partition_cyl_align=on
-boot_image any partition_offset=0
-boot_image any partition_hd_cyl=64
@@ -14,12 +14,6 @@
-boot_image any load_size=2048
-boot_image any boot_info_table=on
-boot_image any next
--boot_image any efi_path='/images/efiboot.img'
--boot_image any platform_id=0xef
--boot_image any emul_type=no_emulation
--boot_image any load_size=11462656
--boot_image isolinux partition_entry=gpt_basdat
--boot_image any next
-boot_image any efi_path='/images/macboot.img'
-boot_image any platform_id=0xef
-boot_image any emul_type=no_emulation
```
```
❯ diff -u ~/xorriso-report-{orig,new5}.log
--- /home/michel/xorriso-report-orig.log 2020-07-15 20:01:01.405289018 -0700
+++ /home/michel/xorriso-report-new5.log 2020-07-15 20:30:27.571644770 -0700
@@ -1,6 +1,6 @@
--volid 'Fedora-S-dvd-x86_64-32'
--volume_date uuid '2020042219432600'
--boot_image isolinux system_area=--interval:imported_iso:0s-15s:zero_mbrpt,zero_gpt:'/home/michel/Downloads/Fedora-Server-netinst-x86_64-32-1.6.iso'
+-volid 'Fedora-32-btrfs'
+-volume_date uuid '2020071603283100'
+-boot_image isolinux system_area=--interval:imported_iso:0s-15s:zero_mbrpt,zero_gpt:'./iso/Fedora-32-WS-20200715-5.iso'
-boot_image any partition_cyl_align=on
-boot_image any partition_offset=0
-boot_image any partition_hd_cyl=64
@@ -17,7 +17,7 @@
-boot_image any efi_path='/images/efiboot.img'
-boot_image any platform_id=0xef
-boot_image any emul_type=no_emulation
--boot_image any load_size=11462656
+-boot_image any load_size=11460608
-boot_image isolinux partition_entry=gpt_basdat
-boot_image any next
-boot_image any efi_path='/images/macboot.img'
```
Signed-off-by: Michel Alexandre Salim <michel@michel-slm.name>
Fails on Fedora 32 and we are going to decommission this test
very soon, leaving only composer-cli related tests here. All image
building and verification tasks are already in the osbuild-composer
test suite!
urllib3 has a retry system that send the request again when the
connection fails, times out, is redircted, etc. This turns off retries,
which aren't very useful anyway.
This is used to start an ostree build, it is only supported on
ostree-composer, and requires the ostree ref and parent. It may also
include --size and optionally be uploaded.
osbuild-composer can support user specified image sizes, this adds an
optional argument, after start, to specify the size in MiB. eg.
composer-cli compose start --size 2048 example-http-server qcow2
This only works when the backend is not 'lorax-composer', when it is the
user will get a warning that it will be ignored.
At least in our CI, the default place to store VM runtime overlays
(testvm.get_temp_dir()) points to a tmpfs file, so that tests can put VM
overlays into RAM and are not affected by slow I/O. But that doesn't
work for composer, as it tends to produce huge overlays due to real-life
OS composed trees.
Use /var/tmp/ instead, which is meant for large files.
Also simplify the VirtMachine invocation -- if `identity_file` is None,
that's fine -- it's the default value of that argument anyway.
This reverts commit 6025da1421.
It ends up that using the install.img with qemu and PXE and fips=1 is a
common use case. Without vmlinuz in the install.img rootfs it has
nothing to run the check against.
Related: rhbz#1782737
Seems petitboot can't properly parse the live image grub config on ppc, thus
booting fails on bare-metal. Fix the problem by removing the obsolete 32-bit
entries.
The host system can run out of space while running mksquashfs, and while
this is logged to program.log it isn't detected by lorax or
livemedia-creator so it will continue running, possibly reporting
unrelated errors and causing confusion.
This adds checks for the return status when calling mksquashfs, logs it
to the log, and either exits or raises an error immediately.
The kernel in /boot is not needed. Keep the .vmlinuz*hmac file so that
fips mode can check it (this requires dracut-050 or later).
Related: rhbz#1782737
Previously the release package was chosen by picking the first package
that provides 'system-release' that isn't named generic. This can cause
branding issues with repos containing multiple system-release packages.
This patch changes _install_branding so that it will give preference to
a system-release package that ends with lowercase --variant name. If
there isn't one it will fall back to the previous behavior.
Resolves: rhbz#1472622
epoch wasn't included in the lorax-packages.log, but arch is. And for
the debuginfo the epoch, including 0, *is* included. This should match
the previous output.
Resolves: rhbz#1815000
A change in glibc now requires /proc be mounted in order to run mknod
which is needed in order to run rpm from runtime-postinstall.
This drops that code from the template and moves writing the package
list into run_transaction, which already has all of the needed
information to generate the list.
Resolves: rhbz#1812895
osbuild-composer also supports the weldr API, and front-ends like
cockpit-composer can use either one of them, so to make it easier to
switch between them we are adding 'weldr' to each of the API servers.
Some of the files no longer exist, some of them have moved. In the case
of dracut the 98systemd directory was renamed to 98dracut-systemd, but
nobody noticed.
This updates the following:
* rename 98systemd to 98dracut-systemd so scripts are in the
install.img
* drop fedora-release removefrom, it now only has os-release
fedora-repos has the repo files, not anaconda, they are moved by
runtime-postinstall.tmpl
* Use initscripts to keep the /etc/init.d, chkconfig only has an empty
directory.
* gtk2-engines is no longer installed
* metacity doesn't include anything in /etc/
* /usr/share/X11/rgb.txt is no longer installed
* libgstbadallocators-1.0.so
The eject utility moved into util-linux and the package was dropped, but
since the runtime-cleanup template is using `removefrom util-linux
--allbut` it was never added to the boot.iso after the move.
This removes the package request for eject and adds it to the list of
binaries to keep from util-linux.
- for keys use Composer-Test-Key prefix (for consistency)
- for images and VMs use Composer-Test prefix
- delete VMs both by searching by tags & by name in case we
manually create them without tags
