QA » openQA Project » openQA Tests

example: https://openqa.opensuse.org/tests/419856

any ETA here? hits us again with current leap build

Hi Ludwig, I just assigned this ticket to me, and am planning to start with it this week. However I am still in progress with another ticket, so please be a bit more patient. Thanks.

The example testrun failed because of zypper_call "in autoyast2" failed. This bug bsc#17436 is already verified fixed. Is it still necessary to seperate yast2_clone_system?

yes, please remove it.

The current yast2_clone_systemy does no more than "zypper in autoyast2" and "yast2 clone_system", I didn't see anywhere about "generating the hard disk image". If it is contained in "yast2 clone_system", it is impossible to separate it.

What should be removed?

the problem is that this test is called from the "gnome-image" test. So if cloning fails the image creation fails which breaks all follow up test. The test that calls yast2_clone_system should therefore be separate and by itself based on "gnome-image" rather than be part of it.

See the link in comment#1. The whole "console" block shouldn't be there.

OK, so you want the whole "console block" (consoletest_setup, yast2_clone_system, consoletest_finish) dropped from gnome-image scenario?

PR created.

https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/3274

So it's as easy as removing CLONE_SYSTEM=1? Then let's remove that from the job template. Note that we need a second job that generates the autoyast profile instead then. Also, the autoyast_reinstall_gnome template needs to be adjusted.

gnome-image is the base for many other test scenarios. I don't think we should simply remove yast2_clone_system and console tests from gnome-image scenario. I'll close my PR.

We don't have many problems with this scenario. We run the same scenario on openSUSE Tumbleweed and SLE. I don't see a strong reason to change the scenarios. Nevertheless I triggered some jobs manually to crosscheck if the test plan would be correct:

$ openqa_clone_job_osd 1054948 CLONE_SYSTEM= _GROUP=0 TEST=create_hdd_gnome
Created job #1057865: sle-12-SP3-Server-DVD-x86_64-Build0472-sles12_gnome_create_hdd@64bit -> https://openqa.suse.de/t1057865
$ openqa_clone_job_osd 1054948 BOOT_HDD_IMAGE=1 PUBLISH_HDD_1= HDD_1=SLES-12-SP3-x86_64-Build0472-gnome.qcow2 CLONE_SYSTEM=1 _GROUP=0 TEST=clone_system START_AFTER_TEST=create_hdd_gnome
Created job #1057868: sle-12-SP3-Server-DVD-x86_64-Build0472-sles12_gnome_create_hdd@64bit -> https://openqa.suse.de/t1057868

The first one creates the HDD, the second one is triggered after that and publishes the cloned profile. The execution of the modules within the same scenario takes about 4 minutes (https://openqa.suse.de/tests/1054948/file/autoinst-log.txt), execution in a standalone job about the same so I'm not concerned about an additional time here though.

Also proposed is to harmonize the scenario names on o3 and osd. I propose to call them create_hdd_… in all cases, e.g. create_hdd_gnome.

Opinions?

As confirmed in today's openQA sync call I am going with create_hdd_…. Renamed all test suites on o3 accordingly (…-image -> create_hdd_…) and added new scenario clone_system triggered in both openSUSE Tumbleweed and openSUSE Leap 42.3 .

Also renamed settings now on osd using dump_templates, local editing with sed and stuff and load_templates with manual crosschecking and deletion of left-over old test suites.

Left to do: Extract the clone_system test suite on osd.

OSD adapted as well.