Raw disk-images and ISO9660 images are created in a Qemu virtual machine. This
is quite fragile, very slow, and almost unusable without KVM.
For all these reasons, add support for host image generation. This implies the
use new image generation mechanisms.
- Raw disk images: images of partitions are created using tools such as mke2fs
and mkdosfs depending on the partition file-system type. The partition
images are then assembled into a final image using genimage.
- ISO9660 images: the ISO root directory is populated within the store. GNU
xorriso is then called on that directory, in the exact same way as this is
done in (gnu build vm) module.
Those mechanisms are built upon the new (gnu image) module.
* gnu/image.scm: New file.
* gnu/system/image.scm: New file.
* gnu/build/image: New file.
* gnu/local.mk: Add them.
* gnu/system/vm.scm (system-disk-image): Rename to system-disk-image-in-vm.
* gnu/ci.scm (qemu-jobs): Adapt to new API.
* gnu/tests/install.scm (run-install): Ditto.
* guix/scripts/system.scm (system-derivation-for-action): Ditto.
This is a follow-up of a860eddbbd. Guided
installation tests are now run from an ISO image. Hence the main block device
is vda and not vdb anymore.
* gnu/tests/install.scm (installation-target-os-for-gui-tests): Use
%minimal-os-on-vda instead of %minimal-os.
(%minimal-os-on-vda): Make sure that it replicates the config of %minimal-os.
This is a follow-up of a860eddbbd. If using an
ISO, the main disk is vda and not vdb anymore.
* gnu/tests/install.scm (installation-target-os-for-gui-tests): Use vda2 as
swap partition.
It seems that 'guix system init' is consuming more than the 800M of RAM
currently allocated. Until this is understood, bump the limit to 1.2G.
Reported here:
https://lists.gnu.org/archive/html/bug-guix/2020-04/msg00519.html
* gnu/tests/install.scm (run-install): Bump RAM to 1.2G.
This commit adjusts the upgrade-services system test to not build anything
when computing the derivation for the system test. I came across this when
looking at issues computing the system test derivations to store in the Guix
Data Service.
* gnu/tests/reconfigure.scm (run-upgrade-services-test): Remove the use of,
and definition for ensure-service-file.
operating-system's kernel.
* gnu/system.scm (package-for-kernel): New procedure.
(operating-system-directory-base-entries): Use it.
* gnu/tests/linux-module.scm: Test it.
Previously we could occasionally try to connect before the server is
actually listening, both for OpenSSH and Dropbear.
* gnu/tests/ssh.scm (run-ssh-test)["wait for port 22"]: New test.
This fixes a regression introduced in 8b9cad01e9
where ovs-vsctl would no longer be available in PATH.
* gnu/tests/networking.scm (run-openvswitch-test): Qualify "ovs-vsctl"
invocation by ungexping the OPENVSWITCH variable instead of assuming it's
available on PATH.
Fixes <https://bugs.gnu.org/25569>.
Reported by Divan Santana <divan@santanas.co.za>.
* gnu/services/base.scm (shepherd-set-http-proxy-action): New procedure.
(guix-shepherd-service): Add 'actions' field. Change 'start' to a
lambda; check the value of the "http_proxy" environment variable and
add "http_proxy" and "https_proxy" to #:environment-variables as a
function of that.
* gnu/tests/base.scm (run-basic-test)["guix-daemon set-http-proxy
action", "guix-daemon set-http-proxy action, clear"]: New tests.
* doc/guix.texi (Base Services): Document it.
Partly fixes <https://bugs.gnu.org/40405>.
Reported by Diego Nicola Barbato <dnbarbato@posteo.de>.
* gnu/services/base.scm (syslog-service-type): Change 'start' method to
set umask to #o137 before spawning syslogd.
* gnu/tests/base.scm (run-basic-test)["/var/log/messages is not
world-readable"]: New test.
* doc/guix.texi (Linux Services): Add a new subsection and document the
new service and its configuration.
* gnu/services/linux.scm (kernel-module-loader-service-type): New type.
(kernel-module-loader-shepherd-service): New procedure.
* gnu/tests/linux-modules.scm (module-loader-program): Procedure
removed.
(modules-loaded?-program): New procedure.
(run-loadable-kernel-modules-test): 'module-loader-program' procedure
replaced by the new one.
[os]: Use 'kernel-module-loader-service'.
Signed-off-by: Danny Milosavljevic <dannym@scratchpost.org>
* gnu/tests/install.scm (gui-test-program): Add a desktop? argument, and pass it to choose-services,
(installation-target-os-for-gui-tests): new procedure,
(installation-target-desktop-os-for-gui-tests): new procedure,
(guided-installation-test): add target-os and desktop? arguments. Use
target-os instead of the previous os variable. Pass desktop? argument to
gui-test-program.
(%test-gui-installed-os): Adapt accordingly,
(%test-gui-installed-os-encrypted): ditto,
(%test-gui-installed-desktop-os-encrypted): new exported variable.
When marionette-eval calls fail in gui-test-program, the installation
continues which results in two scenarios:
- hang forever at the next marionette-eval call,
- keep going and start a broken installation, which is annoying because it
clears the terminal and hides the error.
Make sure that gui-test-program is exited with #f return code when one of the
marionette-eval calls fail.
* gnu/tests/install.scm (gui-test-program): Add a new macro
"marionette-eval*". Throw an exception when one on the marionette-eval calls
fail.
Fixes <https://bugs.gnu.org/39926>.
Reported by Maxim Cournoyer <maxim.cournoyer@gmail.com>.
Previously we'd error out if the installation script exits with
non-zero, which was the case because the 'reboot' program would
typically not get a reply, and thus would eventually be killed by PID 1
as the system is brought down.
* gnu/tests/install.scm (run-install)[install]: Expect SCRIPT to exit
with SIGTERM in addition to exiting with zero.
The test had been failing since the upgrade to 6.6.3p1 in commit
2dbfd8eec4.
* gnu/services/mail.scm (opensmtpd-activation): Create /var/spool/mail.
* gnu/tests/mail.scm (run-opensmtpd-test): Check /var/spool/mail instead
of /var/mail.
Previously the 'wait' loop would run for ~1024 seconds, at which point
we'd reach the file descriptor limit due to the leak in 'queue-empty?'.
* gnu/tests/mail.scm (run-opensmtpd-test)[test]("mail arrived"): In
'queue-empty?', close PIPE to avoid file descriptor leak. In 'wait'
loop, arrange to run at most 20 times.
This is a followup to 8b9cad01e9, which
would leave PATH unset.
* gnu/tests/rsync.scm (run-rsync-test)[test]("service running"): Add
call to 'setenv' for PATH.
* gnu/tests/mail.scm (%getmail-os): Rewrite so that the "alice" account
has a password.
(run-getmail-test)[test]("set password for alice"): Remove. This would
not work since commit 8b9cad01e9 since
'passwd' would no longer be in $PATH.
The system image would no longer fit in memory, starting from commit
8c7eb58453.
* gnu/tests/docker.scm (run-docker-system-test): Set 'memory-size' to 3500.
* gnu/tests/install.scm (%test-gui-installed-os-encrypted): New variable,
(guided-installation-test): set a swap-device only if there is no encryption.
This follows up on commit 0d48690908.
* gnu/services/mail.scm (%default-opensmtpd-config-file): Adapt to ‘new’
≥6.4 grammar.
* gnu/tests/mail.scm (%opensmtpd-os): Likewise.
Fixes <https://bugs.gnu.org/38086>.
Thanks to Vagrant and Tobias!
* gnu/tests/install.scm (%raid-root-os)[initrd-modules]: Add "raid1"
instead of "raid0".
(%raid-root-installation-script): Make the partitions twice as big.
Invoke 'mdadm' with '--level=mirror' instead of '--level=stripe';
connect "yes" to its stdin.
(%test-raid-root-os): Set #:target-size to 2.8 GiB.
1.2G had become slightly too small on x86_64.
This is a followup to 8dfb0c969e.
* gnu/tests/install.scm (%simple-installation-script)
(%extlinux-gpt-installation-script)
(%simple-installation-script-for-/dev/vda): Switch from 1.2G to 1.4G.
* gnu/services/guix.scm: New file.
* gnu/tests/guix.scm: New file.
* gnu/local.mk (GNU_SYSTEM_MODULES): Add both new files.
* doc/guix.texi (Guix Services): New section documenting the Guix Data
Service.
Fixes <https://bugs.gnu.org/37161>.
* guix/scripts/pack.scm (docker-image)[build]: Add a 'directory' entry
for "/tmp" to DIRECTIVES.
* tests/pack.scm ("docker-image + localstatedir"): Test the presence of /tmp.
* gnu/tests/docker.scm (run-docker-test)["Load docker image and run
it"]: Test the presence and permission bits of "/tmp".
* gnu/tests/reconfigure.scm (run-switch-to-system-test): Assert that
'/run/current-system' points to the activated system, and that new user
accounts specified in the operating system declaration are created.
[[PGP Signed Part:No public key for F506FD169D8FD95A created at 2019-07-24T18:35:10+0200 using RSA]]
* gnu/tests/reconfigure.scm: New file.
* gnu/local.mk (GNU_SYSTEM_MODULES): Add it.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
* guix/scripts/pack.scm (singularity-environment-file): New procedure.
(squashfs-image): Use it, and create /.singularity/env/90-environment.sh.
* gnu/tests/singularity.scm (run-singularity-test)["singularity run,
with environment"]: New test, currently skipped.
* gnu/tests/singularity.scm (build-tarball&run-singularity-test): Add
GUILE-JSON to the profile.
* guix/docker.scm (config): Add #:environment parameter and honor it.
(build-docker-image): Likewise, and pass it to 'config'.
* guix/scripts/pack.scm (docker-image): Import (guix profiles) and (guix
search-paths). Call 'profile-search-paths' and pass #:environment to
'build-docker-image'.
* gnu/tests/docker.scm (run-docker-test)["Load docker image and run it"]:
Add example that expects (json) to be available.
* gnu/tests/docker.scm (build-tarball&run-docker-test): Replace
%BOOTSTRAP-GUILE by GUILE-2.2 and GUILE-JSON in the environment.
