QA » openQA Project » openQA Tests

Good idea.

as reported in #33583 we still see errors which seem to be hyperv-timeout related.

@michalnowak what would you say, increase TIMEOUT_SCALE even more or is it something else or yast2_i specific timeout increase?

okurz wrote:

as reported in #33583 we still see errors which seem to be hyperv-timeout related.

@michalnowak what would you say, increase TIMEOUT_SCALE even more or is it something else or yast2_i specific timeout increase?

Actually the subject of that ticket was wrong to blame Hyper-V, it was Xen PV. It should not happen there at all, will investigate it as it happened for the second time in two days.

text-login-20180315 is prematurely matching on TTY1 while we are waiting for the console to be switched to TTY6. This is where the switch is still in progress but openQA start typing login name: https://openqa.suse.de/tests/1607827#step/consoletest_setup/3.

In situations where Hyper-V is faster, we are luckier and both text-login-20180315 and tty6-selected are matched: https://openqa.suse.de/tests/1605086#step/consoletest_setup/2 But only the latter one should have been.

https://gitlab.suse.de/openqa/os-autoinst-needles-sles/merge_requests/809 to remove needle. Will propose option to match "tty$nr-selected", currently we have

if (match_has_tag("tty$nr-selected") or match_has_tag("text-login")) {
     type_string "$user\n";
     handle_password_prompt;
}

https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/4891

wait_still_screen 3 didn't help here: https://openqa.opensuse.org/tests/659550#step/zypper_add_repos/1
Proposing: https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/4907 to identify all problematic cases and then decide how to proceed.

So, remaining part is ti get stats when we get soft-failure, create required needles and after that assert tty by default.

Taking the interval 11 days, since the pull request creating the soft-failure, searching poo#32926 for SLE we can found the soft-failure only in test mau-qa_kernel_kexec:

• sle 15 & sle-12-SP3 for aarch64 & x86_64 all soft-failing since 14 days ago, except for 1 passed out 10 for sle 15 x86_64. Searching for openSUSE there are more test suites affected:
• Only for aarch64 in openSUSE Tumbleweed AArch64. Available statistics 1 out 1, soft-failing in create_hdd_gnome, create_hdd_textmode, minimalx, textmode. All needles seems have been already created in both side.

still found references of the soft-fail, e.g. https://openqa.opensuse.org/tests/674191?#step/NM_wpa2_enterprise/31 where the soft-fail refers to a needle needed for the post_fail_hook. Albeit, if this test case would fail early because the tty5-selected needle would not exist it is not a big problem, one would create the needle and rerun the test to execute the post_fail_hook properly. I created a tty5-selected needle and retriggered

The area was missing, once I created the area for tty5 and restarted is not displaying the soft-failure anymore: https://openqa.opensuse.org/tests/674288#

Any progress here? I would at least merge PR from Joaquin and then you can improve solution with your proposal. WDYT?

So I did the corresponding steps for o3 as well.

https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/5023 was the PR from jrivera which we already closed.

https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/5026 is my PR

Looking at

https://w3.nue.suse.com/~okurz/openqa_opensuse_org_status.html I can find no references to poo#32926, however on https://w3.nue.suse.com/~okurz/openqa_suse_de_status.html I can still find references to "kernel_kexec" that look for "linux-login" and make it wrong as the prompt is looked for before actual reboot which I think we need to handle first.

https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/5057 first, then we should be done and go with https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/5026

have not found any related issues checking https://w3.nue.suse.com/~okurz/openqa_opensuse_org_status.html and https://w3.nue.suse.com/~okurz/openqa_suse_de_status.html today