Add jobs to build and publish update bundles for qemu - both x86 and
x86_64

Signed-off-by: Stevan Radaković <stevan.radakovic@linaro.org>

When looking at pipeline [1] we have realized that the newly introduced
bundle build and publishing jobs pose a problem for unrelated updates
of the tree, as depending on which exact files are modified by the
incoming pull request, the build jobs may not spawn, causing a
dependency error on the bundle and publishing jobs.

When initially introduced, it was overlooked that the jobs both have a
dependency on one another (build->bundle->publish) *and* come with a
distinct set of rules that govern if a specific job is created in a
given pipeline.

To resolve the problem move the rules to a new dot-job, called
.workspace-rules, and inject the dependency on .workspace-rules to all
the jobs that share the dependency chain.

[1] https://gitlab.eclipse.org/eclipse/oniro-core/oniro/-/pipelines/4285



Signed-off-by: Zygmunt Krynicki <zygmunt.krynicki@huawei.com>

Chase Qi authored 2 years ago and Stevan Radaković committed 2 years ago

Signed-off-by: Chase Qi <chase.qi@linaro.org>

Signed-off-by: Zygmunt Krynicki <zygmunt.krynicki@huawei.com>

This expands automatic updates to the SECO B68 board. Both new jobs
re-use existing generic jobs and contain no new logic.

Signed-off-by: Zygmunt Krynicki <zygmunt.krynicki@huawei.com>

The HawkBit instance needs to be configured externally and reachable
from the runner performing the publishing step. The location of the
instance must be provided through environment variables compatible with
hawkbitctl, namely HAWKBIT_{URL,USERNAME,PASSWORD}. HAWKBIT_TENANT is
optional and untested.

The bundle is added as an artifact to a new software module named after
the bundle, using bundle timestamp as the version. The module is then
placed in a distribution named after the image (not the bundle), with
the exact same version.

Signed-off-by: Zygmunt Krynicki <zygmunt.krynicki@huawei.com>

This allows the bundle job to be started even if just one of the jobs
in the previous stage, the one for pi4 image build, succeeds. This is
better for testing and allows jobs to start and finish faster, reducing
total build time.

Signed-off-by: Zygmunt Krynicki <zygmunt.krynicki@huawei.com>

Enable smoke, ltp, kselftest and perf tests on seco-c61.

Signed-off-by: Chase Qi <chase.qi@linaro.org>

Chase Qi authored 2 years ago and Andrei Gherzan committed 2 years ago

Signed-off-by: Chase Qi <chase.qi@linaro.org>

The EF infrastructure (GitLab) doesn't support image registry. This
moves the CI logic of building and deploying all the container images to
the OSTC instance while EF catches up[1].

[1] eclipsefdn/helpdesk#1108



Signed-off-by: Andrei Gherzan <andrei.gherzan@huawei.com>

Chase Qi authored 3 years ago and Stevan Radaković committed 3 years ago

Signed-off-by: Chase Qi <chase.qi@linaro.org>

Signed-off-by: Chase Qi <chase.qi@linaro.org>

Set lava-test and lava-report job as scheduled pipeline only for the
below reasons.

* lava-test requires token set using GitLab variable for job submission
* lava-report depends on all test jobs

Signed-off-by: Chase Qi <chase.qi@linaro.org>

Signed-off-by: Chase Qi <chase.qi@linaro.org>

Zygmunt Krynicki authored 3 years ago and Stefan Schmidt committed 3 years ago

The update bundle is built with the new hidden job .build-rauc-bundle.
The job is documented and instantiated in the main pipeline.

Signed-off-by: Zygmunt Krynicki <zygmunt.krynicki@huawei.com>

Signed-off-by: Chase Qi <chase.qi@linaro.org>

Signed-off-by: Andrei Gherzan <andrei.gherzan@huawei.com>

Signed-off-by: Chase Qi <chase.qi@linaro.org>

The 'when: never' rule should be put on the top to prevent gitlab from
adding the job to pipeline.

Signed-off-by: Chase Qi <chase.qi@linaro.org>

Chase Qi authored 3 years ago and Stevan Radaković committed 3 years ago

Signed-off-by: Chase Qi <chase.qi@linaro.org>

This closes: #226

Signed-off-by: Chase Qi <chase.qi@linaro.org>

lava-test badge shows success rate for on device testing in lava.

Signed-off-by: Chase Qi <chase.qi@linaro.org>

Introduce machine variable for lava-test and search build images for the
test target device only to make sure only images for the target device
submitted for test.

This closes #215.

Signed-off-by: Chase Qi <chase.qi@linaro.org>

Signed-off-by: Chase Qi <chase.qi@linaro.org>

Chase Qi authored 3 years ago and Stevan Radaković committed 3 years ago

Signed-off-by: Chase Qi <chase.qi@linaro.org>

After all test jobs for a pipeline finish, trigger lava-report via gitlab api
to collect test results from all jobs.

Signed-off-by: Chase Qi <chase.qi@linaro.org>

* fix the syntax for CI/CD var check
* enable lava report

Signed-off-by: Chase Qi <chase.qi@linaro.org>

CI_LAVA_TOKEN is required for submitting job to LAVA server. It should
be set in instance level gitlab CI/CD variables for testing forking dev
flow based MRs.

With `when: never` rule, test job will not be added to pipeline if the
above var not set. If the condition not set in test job that extends
`.lava-test`, `.lava-test` will ask user to do so and then exit with
unsuccessful job status. There is no easy way to exit early with
successful status yet. See
https://gitlab.com/gitlab-org/gitlab/-/issues/329217#note_572963049



Adding the above condition in `.lava-test` has no impact on test job
that extends the `.lava-test`.

Signed-off-by: Chase Qi <chase.qi@linaro.org>

Signed-off-by: Chase Qi <chase.qi@linaro.org>

Signed-off-by: Andrei Gherzan <andrei.gherzan@huawei.com>

This may be reverted ASAP,
this change was suggested by @zyga to suspend bot failures
on release

Cc: @zyga
Forwarded: https://booting.oniroproject.org/distro/oniro/-/merge_requests/383


Signed-off-by: Philippe Coval <philippe.coval@huawei.com>

Zygmunt Krynicki authored 3 years ago and Stefan Schmidt committed 3 years ago

Changes to the pipeline definition file should trigger the same set of
builds as changes to the shared jobs.

Signed-off-by: Zygmunt Krynicki <zygmunt.krynicki@huawei.com>

Zygmunt Krynicki authored 3 years ago and Stefan Schmidt committed 3 years ago

The manifest is special, because when it changes the oniro repository is
cloned three times inside the CI process:

- first time by GitLab, as a part of the normal CI checkout process
- second time by repo init, as a part of the .workspace job
- last time by repo sync, as a part of the same job

When the manifest changes, the CI system should be testing the incoming
change. Crucially the second step needs to be adjusted, to take the
manifest from a non-default location. This was handled before the
manifest repository was merged into the oniro repository. During that
merge the critical step was lost.

Adjust the .bitbake-workspace job to inject variables that pick the
correct manifest repository and branch. Leave a comment for the untested
rule about CI for tags.

Signed-off-by: Zygmunt Krynicki <zygmunt.krynicki@huawei.com>

REUSE is allowed to fail, DCO is enforced.

Signed-off-by: Zygmunt Krynicki <zygmunt.krynicki@huawei.com>

Signed-off-by: Zygmunt Krynicki <zygmunt.krynicki@huawei.com>

The CI system depends on a number of containers. Those used to be
defined in another repository but with this migration we can now
simplify and aggregate everything into one repository.

Signed-off-by: Zygmunt Krynicki <zygmunt.krynicki@huawei.com>

This hidden job allows building Docker containers without privileged or
nested docker. It will be used to build all the containers used in CI.

Signed-off-by: Zygmunt Krynicki <zygmunt.krynicki@huawei.com>

With the move to booting.oniroproject.org, the two compliance jobs now
refer to non-existent repository. They will be re-enabled soon, after
all containers and shared jobs definitions are merged into the oniro
project.

Signed-off-by: Zygmunt Krynicki <zygmunt.krynicki@huawei.com>

This template is offered by GitLab to enable pipelines for merge
requests, a safety feature that avoids breaking the default branch by
merging things that passed in the past.

Reference: https://docs.gitlab.com/ee/ci/pipelines/merge_request_pipelines.html


Signed-off-by: Zygmunt Krynicki <zygmunt.krynicki@huawei.com>

The machines-and-flavours.yaml file needs to be included after
build-generic.yaml. Order matters because of the delicate manner
in which they override and extend jobs seen previously.

Signed-off-by: Zygmunt Krynicki <zygmunt.krynicki@huawei.com>