This brings all tests runtime combined from ~40 minutes down to 4 minutes
when I run it locally on my laptop. In gitlab pipeline it goes down
from usual 15 minutes to ~6.

Couple tests that depend on yaml validation, enable it just for
duration of those tests.

Signed-off-by: Jan Stancek <jstancek@redhat.com>

There's up to 100 discussions in each loop, so 5 repetitions of the
test should be enough - given that we run these tests for each MR
and also locally.

Signed-off-by: Jan Stancek <jstancek@redhat.com>

See merge request !1743

Part of: &123



Signed-off-by: Tales da Aparecida <tales.aparecida@redhat.com>

See merge request !1742

There's up to 100 discussions in each loop, so 5 repetitions of the
test should be enough - given that we run these tests for each MR
and also locally.

Signed-off-by: Jan Stancek <jstancek@redhat.com>

This was missed from !1632.

Signed-off-by: Patrick Talbert <ptalbert@redhat.com>

See merge request !1741

Part of: &123



Signed-off-by: Tales da Aparecida <tales.aparecida@redhat.com>

There are 2 issues:

1. Mock says that mock_ji has alias attribute, because it says
so for any attribute you check. When webhook code tries to iterate
over CVEs, it finds another Mock object, and it fails at:
  File "/home/jstancek/src/kernel-workflow-git/webhook/rhissue.py", line 465, in is_cve_tracker
    return bool(self.cve_ids)
                ^^^^^^^^^^^^
  File "/home/jstancek/src/kernel-workflow-git/webhook/rhissue.py", line 428, in cve_ids
    return self._get_cve_ids(self.ji.alias) \
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jstancek/src/kernel-workflow-git/webhook/rhissue.py", line 591, in _get_cve_ids
    for string in string_list:
                  ^^^^^^^^^^^
  TypeError: 'Mock' object is not iterable

2. fixVersion is initialized as MagicMock(name='...'), which doesn't
create and set name attribute. So when code tries to use it,
it finds another MagicMock object and fails at:
  File "/home/jstancek/src/kernel-workflow-git/webhook/rhissue.py", line 566, in _find_ji_project_branch
    if not self.ji or not self.ji_fix_version:
                          ^^^^^^^^^^^^^^^^^^^
  File "/home/jstancek/src/kernel-workflow-git/webhook/rhissue.py", line 246, in ji_fix_version
    return defs.FixVersion(fix_versions[0].name)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jstancek/src/kernel-workflow/venv_webhooks/lib64/python3.12/site-packages/kwf_lib/common.py", line 102, in __init__
    if fv_match := CENTOS_FIX_VERSION_REGEX.match(fv_string):
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  TypeError: expected string or bytes-like object, got 'MagicMock'

Use already created FakeJI instead of Mock and initialize fixVersion name attribute explicitly.

Signed-off-by: Jan Stancek <jstancek@redhat.com>

See merge request !1740

This speeds up tests and ack_nack webhook, because owners isn't reloaded all the time.

Signed-off-by: Jan Stancek <jstancek@redhat.com>

See merge request !1736

This was missed from !1632.

Signed-off-by: Patrick Talbert <ptalbert@redhat.com>

There are 2 issues:

1. Mock says that mock_ji has alias attribute, because it says
so for any attribute you check. When webhook code tries to iterate
over CVEs, it finds another Mock object, and it fails at:
  File "/home/jstancek/src/kernel-workflow-git/webhook/rhissue.py", line 465, in is_cve_tracker
    return bool(self.cve_ids)
                ^^^^^^^^^^^^
  File "/home/jstancek/src/kernel-workflow-git/webhook/rhissue.py", line 428, in cve_ids
    return self._get_cve_ids(self.ji.alias) \
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jstancek/src/kernel-workflow-git/webhook/rhissue.py", line 591, in _get_cve_ids
    for string in string_list:
                  ^^^^^^^^^^^
  TypeError: 'Mock' object is not iterable

2. fixVersion is initialized as MagicMock(name='...'), which doesn't
create and set name attribute. So when code tries to use it,
it finds another MagicMock object and fails at:
  File "/home/jstancek/src/kernel-workflow-git/webhook/rhissue.py", line 566, in _find_ji_project_branch
    if not self.ji or not self.ji_fix_version:
                          ^^^^^^^^^^^^^^^^^^^
  File "/home/jstancek/src/kernel-workflow-git/webhook/rhissue.py", line 246, in ji_fix_version
    return defs.FixVersion(fix_versions[0].name)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/jstancek/src/kernel-workflow/venv_webhooks/lib64/python3.12/site-packages/kwf_lib/common.py", line 102, in __init__
    if fv_match := CENTOS_FIX_VERSION_REGEX.match(fv_string):
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  TypeError: expected string or bytes-like object, got 'MagicMock'

Use already created FakeJI instead of Mock and initialize fixVersion name attribute explicitly.

Signed-off-by: Jan Stancek <jstancek@redhat.com>

This speeds up tests and ack_nack webhook, because owners isn't reloaded all the time.

Signed-off-by: Jan Stancek <jstancek@redhat.com>

Today is the day we rename cki-jira to kwf-lib.

This MR just updates that to reflect the change.

Edits: 
- v2-v5: Just fixing `isort` mess.
- v6: Change to use the official name and main branch (it is not temporary anymore).

Signed-off-by: Julio Faracco <jfaracco@redhat.com>

See merge request !1726

The SSG field gets populated based on the default component assignee, and
when we override the component assignee to SE, we also have to make sure
we clear the SSG field, or you wind up with some confusing assignment and
SSG combinations.

Fixes: #817  

Signed-off-by: Jarod Wilson <jarod@redhat.com>

Closes #817

See merge request !1731

Signed-off-by: Jarod Wilson <jarod@redhat.com>

See merge request !1732

Use them for all lead_stream streams, not just y-stream, since they should
be like this for terminal z-streams as well.

And let's do the indenting in the MR description as well as commits.

Signed-off-by: Jarod Wilson <jarod@redhat.com>

See merge request !1733

Make life easier for some automated parsers, and indent the upstream
commit data in the MR description too, not just in the commits.

Signed-off-by: Jarod Wilson <jarod@redhat.com>

Use them for all lead_stream streams, not just y-stream, since they should
be like this for terminal z-streams as well.

Signed-off-by: Jarod Wilson <jarod@redhat.com>

This commit just replace the cki_jira module name to kwf_lib.

Signed-off-by: Julio Faracco <jfaracco@redhat.com>

This commit changes setup.cfg to use the new location of cki-jira
library. Now, it is called kwf-lib and we should reflect that change in
setup.cfg.

Signed-off-by: Julio Faracco <jfaracco@redhat.com>

Signed-off-by: Jarod Wilson <jarod@redhat.com>

Now with podman/toolbox-based optional Step 0 commands to get a
Fedora 40 container to use as a working base, for those running
some other Linux system with podman on it. (Tested with both a
CentOS Stream 9 install and macOS with podman, though macOS
requires a podman machine ssh first to run toolbox commands.)

Signed-off-by: Jarod Wilson <jarod@redhat.com>

See merge request !1729

Signed-off-by: Jarod Wilson <jarod@redhat.com>

The SSG field gets populated based on the default component assignee, and
when we override the component assignee to SE, we also have to make sure
we clear the SSG field, or you wind up with some confusing assignment and
SSG combinations.

Fixes: #817

Signed-off-by: Jarod Wilson <jarod@redhat.com>

We need to be able to access SSG, as it needs to be cleared in some bug
reassignment cases.

Signed-off-by: Jarod Wilson <jarod@redhat.com>

It appears a specific versioned layout may require a slightly more complex
setup, involving pyenv, but if your python version is in the right
ballpark, then just 'layout python3' should work fine.

Signed-off-by: Jarod Wilson <jarod@redhat.com>

Signed-off-by: Jarod Wilson <jarod@redhat.com>

This adds some basic steps to install a Fedora 41-based toolbox container
on top of podman for those on other setups that would like a devel env as
close to the current production hook envs as possible. Tested locally with
a clean CentOS Stream 9 install.

While I was here, added some additional packages to the install list that
are required to get through the initial tox install and a full tox
self-test run.

Also noted that the initial clone assumes you've set up an ssh public key
with gitlab for cloning.

Signed-off-by: Jarod Wilson <jarod@redhat.com>

Signed-off-by: Jarod Wilson <jarod@redhat.com>

This reverts commit e3659829.

We are close to some deadlines for c10s and c9s and should avoid the
ckihook demanding MRs grow new pipeline results.

Part of: #808

Signed-off-by: Patrick Talbert <ptalbert@redhat.com>

See merge request !1730

This reverts commit e3659829.

We are close to some deadlines for c10s and c9s and should avoid the
ckihook demanding MRs grow new pipeline results.

Signed-off-by: Patrick Talbert <ptalbert@redhat.com>

The cron job searches for things to prod uses a project-wide graphql query
that we need to do some filtering on to omit things we don't want to look
at. We could try to feed in the list of valid branches to the query, which
might mean fetching an MR or two less from gitlab, but a fair bit of added
code, or we can just do this, and filter out any MR that doesn't map to a
supported/known rh_branch after fetching and building MergeMR objects.

Signed-off-by: Jarod Wilson <jarod@redhat.com>

See merge request !1728

The cron job searches for things to prod uses a project-wide graphql query
that we need to do some filtering on to omit things we don't want to look
at. We could try to feed in the list of valid branches to the query, which
might mean fetching an MR or two less from gitlab, but a fair bit of added
code, or we can just do this, and filter out any MR that doesn't map to a
supported/known rh_branch after fetching and building MergeMR objects.

Signed-off-by: Jarod Wilson <jarod@redhat.com>

Part of #808

See merge request !1727

Part of #808



Signed-off-by: Michael Hofmann <mhofmann@redhat.com>

The merge conflicts check cron job was falling down on kernel-ark's
feodra-6.13 merge branch because it doesn't map to an rh-branch in our
metadata. Handle it gracefully.

Signed-off-by: Jarod Wilson <jarod@redhat.com>

See merge request !1725

The merge conflicts check cron job was falling down on kernel-ark's
feodra-6.13 merge branch because it doesn't map to an rh-branch in our
metadata. Handle it gracefully.

Signed-off-by: Jarod Wilson <jarod@redhat.com>

This is multiple little fixes all rolled into one.

First up, make the revert check handle an entry w/o a subject. Second,
make the old non-kerneloscope find_already_backported_orig function
actually include subject data. Third, update cs10's metadata to use
kerneloscope now, since cs10 data is available there now.

Any of these alone would fix a current crash we're seeing, but all three
seem worth doing, for robustness, consistence, etc.

Signed-off-by: Jarod Wilson <jarod@redhat.com>

See merge request !1724

Since the last time we updated upstream kernel git tree entries, the
wireless tree has moved, and a new wireless-next tree has spawned. Both of
these are trees imported to kerneloscope, so we need to have the refs as
well, so we can access the full git commits if they're found in kerneloscope
queries. Yes, this is a little fragile, and maybe we should just pull
commits entirely from kerneloscope somehow, but ... not today.

Signed-off-by: Jarod Wilson <jarod@redhat.com>

See merge request !1723

We have this filtering for task creation, but also need it for
generation of the test_variants list, so that the missing tasks test
actually passes on things like streams w/o RT even on subsystems that
would propose RT testing variants.

Fixes: #746

Signed-off-by: Jarod Wilson <jarod@redhat.com>

Closes #746

See merge request !1722

We need to let libbackport assemble the backported patch list, so
CVE-based filtering works properly, and an extra line return is needed so
the output of the first comment about reviewers looks correct.

Signed-off-by: Jarod Wilson <jarod@redhat.com>

See merge request !1719