action #153577
closedcoordination #151984: [epic] Unify YaST and Migration CIs and job groups
Enhance migration from Leap to SLE
0%
Description
Migrations from Leap to SLE always fail before RC:
https://openqa.suse.de/tests/overview?distri=sle&version=15-SP5&build=80.5&groupid=478
and due to we want to reduce ambiguity of softfailures we shouldn't even mark them as such (related with epic #151822).
Therefore what we could do is to put an exception in the code that:
- If this particular product is not found in SCC, https://openqa.suse.de/tests/13095371#step/register_system/12 (perhaps we need to grap the logs?) we set the status of the job as passed unless development phase is RC or GMC, where we should normally fail and have a proper bug to label the job.
We are not going to use too many flavor due they create a cost of setup unnecessary with a good renaming and re-structuring of our test suites.
In order to move test suite from one flavor to another we will need need reconfiguration of settings in job group yaml.
Acceptance criteria¶
AC1: Move these three jobs to Migration job group (to flavor Migration-from-SLE15-SPx to be deleted in the future as well) without using its existing flavor.
AC2: Apply proper naming that fits well with existing new naming (discuss with the squad in Slack for feedback).
AC3: In order to get the latest Leap, apply the best solution as result of #125549, I guess option 1, because 3 not sure it could work out of the box (contact with @tinawang123 in any doubt)
AC4: Add the logic necessary to pass the job always before RC phase, while fail normally from there.
Additional information¶
Jobs to move: https://openqa.suse.de/tests/overview?arch=&flavor=Migration-from-Leap-15.x-to-SLE15-SPx&machine=&test=&modules=&module_re=&group_glob=¬_group_glob=&distri=sle&version=15-SP6&build=45.1&groupid=478#
For condition of development phase, please see https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/18441
Updated by JERiveraMoya about 1 year ago
- Description updated (diff)
- Parent task changed from #151990 to #151984
Updated by JERiveraMoya about 1 year ago
- Tags changed from qe-yam-jan-sprint to qe-yam-feb-sprint
Updated by JERiveraMoya about 1 year ago
- Tags changed from qe-yam-feb-sprint to qe-yam-mar-sprint
Updated by tinawang123 about 1 year ago
- Status changed from Workable to In Progress
- Assignee set to tinawang123
Updated by tinawang123 about 1 year ago
Updated by JERiveraMoya about 1 year ago
please provide PRs/MRs to each ticket to make easy the review and navigation.
Updated by tinawang123 about 1 year ago
Should we wait this bug fix then put the jobs to the migration daily job group?
If don't, I will open one MR.
Updated by JERiveraMoya about 1 year ago
The easy way is to have automation failing in the bug -> it would also help to not forget to include it in the report.
Job runs in development shouldn't be used to file bugs (they are much less stable for developers to take a look).
I would suggest you to focus in small increments so it would be easier also to provide you with feedback, code review or other actions.
In general PRs/MRs in draft are usually a MUST to understand VRs.
Updated by JERiveraMoya about 1 year ago
Still the goal of this ticket is not having those test suite failing all the time before RC, it is expected you find different bugs.
Updated by JERiveraMoya about 1 year ago
Could you please open MR to keep reviewing work here? thanks.
Updated by tinawang123 about 1 year ago
Opened one MR: https://gitlab.suse.de/qe-yam/openqa-job-groups/-/merge_requests/107
There is one problem to use 'HDD_1_URL' to the leap migration to sle job.
Eg: HDD_1_URL: 'https://openqa.opensuse.org/tests/3977685/asset/hdd/opensuse-15.6-x86_64-623.2-kde@64bit.qcow2'
This setting need update if the job '3977685' is deleted.
Updated by JERiveraMoya about 1 year ago
tinawang123 wrote in #note-11:
Opened one MR: https://gitlab.suse.de/qe-yam/openqa-job-groups/-/merge_requests/107
There is one problem to use 'HDD_1_URL' to the leap migration to sle job.
Eg: HDD_1_URL: 'https://openqa.opensuse.org/tests/3977685/asset/hdd/opensuse-15.6-x86_64-623.2-kde@64bit.qcow2'
This setting need update if the job '3977685' is deleted.
We need a new ticket to implement in our CI a query to get the latest id and set the setting without manual intervention.
could you please create a ticket for that?
Updated by tinawang123 12 months ago
Opened new ticket: https://progress.opensuse.org/issues/157195
Updated by JERiveraMoya 12 months ago
what about AC4? do you have any proposal?
Updated by tinawang123 12 months ago · Edited
I did one workaround to ignore this conflict issue during the yast migration.
Refer job: https://openqa.suse.de/tests/13809133#step/yast2_migration/9
Do we need set it softfailed?
Updated by JERiveraMoya 12 months ago · Edited
could you add here the final job runs once you add SCC, in this ticket I would also like to achieve AC4, but now we know that instead of RC, should be GMC and make it green before that.
Updated by JERiveraMoya 12 months ago · Edited
JERiveraMoya wrote in #note-17:
could you add here the final job runs once you add SCC, in this ticket I would also like to achieve AC4, but now we know that instead of RC, should be GMC and make it green before that.
I'm actually fine if we don't do AC4, it is still unclear tbh, if arrives in GMC, it might not be time to adjust anything, so having failing all the time could make sense until we understand better phub.
I meant in the last paragraph proxy SCC (my bad!) and in slack also I mistyped! uups!
Slack conversation: https://suse.slack.com/archives/C02CLB2LB7Z/p1710931565623169?thread_ts=1710929769.443009&cid=C02CLB2LB7Z
Updated by JERiveraMoya 12 months ago
- Status changed from In Progress to Resolved