DRAFT: SLE-M: introduced the ownership filter #20797
Conversation
m-dati
requested review from
mloviska,
jlausuch,
ricardobranco777 and
rbmarliere
as code owners
|
Great PR! Please pay attention to the following items before merging: Files matching
This is an automatically generated QA checklist based on modified files. |
m-dati
changed the title
|
Note: having test slem_image_default_test added with MR 1974, after this PR merged, we can add the QEC parameter filter too in the yaml, to run owned sle-m tests only. |
lib/main_micro_alp.pm
Outdated
| @@ -32,6 +32,11 @@ sub is_dvd { | |||
| return get_required_var('FLAVOR') =~ /dvd/i; | |||
| } | |||
|
|
|||
| sub is_qec_test_run { | |||
| my $vx = get_var('QEC_TEST_FILTER'); | |||
There was a problem hiding this comment.
| my $vx = get_var('QEC_TEST_FILTER'); | |
| return check_var('SLEM_IMAGE_TESTS', '1'); |
QEC_TEST_FILTER appears too generic to me
There was a problem hiding this comment.
OK; but being it dedicated to ownership, then better: QEC_SLEM_IMAGE_TESTS
There was a problem hiding this comment.
Also, I considered the case when param. undefined no filtering shall be applied
There was a problem hiding this comment.
But, another consideration on name: this new param. is not only for image tests, even being now used for that case only 🙂, so I'd prefer to call it QEC_SLEM_TESTS or QEC_SLEM_TESTS_FILTER
There was a problem hiding this comment.
QEC runs container engine, SLEM image and BCI tests on SLE-Micro. We have already now three different test cases, so I'd argue that QEC_SLEM_TESTS or QEC_SLEM_TESTS_FILTER does not distinguish between them sufficiently.
m-dati
force-pushed
the
modules_filter
branch
from
480fe47 to
860b077
Compare
|
VRs rerun, after review applied, https://openqa.suse.de/tests/16185952 VR2 https://openqa.suse.de/tests/16185953 containers: non regression
|
When =1 ownership triggered, based on poo#161150 and coverage table
m-dati
force-pushed
the
modules_filter
branch
from
860b077 to
e409c24
Compare
|
Although this is one of the possible solutions, I would personally prefer to have a set of functions that are loaded according to test definition. This way we avoid creating team specific spaghetti and give it a more logical definition. In TW there is A code example would be over here https://github.com/os-autoinst/os-autoinst-distri-opensuse/blob/master/lib/main_micro_alp.pm#L438 In our case, we need to load only tests that are testing the image setup itself. So we can either create a several atomic load functions, and then specify them using |
|
Here are some VR cloned from last added test Adding the |
m-dati
changed the title
|
Given the discussion still ongoing, PR set draft, for new analysis. |
Introduced, for SLE-Micro test code flow, the
QEC_SLEM_TESTS_FILTERparameter:when =1, the ownership is triggered, based on poo#161150 and coverage table, so that Not qe-c owned test modules are Skipped.
Related ticket: https://progress.opensuse.org/issues/161150
Verification run, set
QEC_TEST_FILTER=1:https://openqa.suse.de/tests/16185796 VR1
https://openqa.suse.de/tests/16185818 non regression
See newer runs below with
QEC_SLEM_TESTS_FILTER.