Re: [PATCH 2/2] tests/avocado/boot_linux.py: re-enable test-case for ppc64

[Alex Bennée]

Kautuk Consul <kconsul@linux.vnet.ibm.com> writes:

> Fixes c0c8687ef0("tests/avocado: disable BootLinuxPPC64 test in CI").
>
> Commit c0c8687ef0fd990db8db1655a8a6c5a5e35dd4bb disabled the test-case
> for PPC64. On investigation, this turns out to be an issue with the
> time taken for downloading the Fedora 31 qcow2 image being included
> within the test-case timeout.
> Re-enable this test-case by setting the timeout to 360 seconds just
> before launching the downloaded VM image.
>
> Signed-off-by: Kautuk Consul <kconsul@linux.vnet.ibm.com>
> Reported-by: Alex Bennée <alex.bennee@linaro.org>
> Tested-by: Hariharan T S hariharan.ts@linux.vnet.ibm.com

It doesn't really address the principle problem that the
boot_linux.py:BootLinuxPPC64.test_pseries_tcg is super heavyweight for
only 2% extra coverage of the executed lines.

What we really need is a script so we can compare the output between the
two jsons:

  gcovr --json --exclude-unreachable-branches --print-summary -o coverage.json 
--root ../../ . *.p

because I suspect we could make up that missing few % noodling the
baseline test a bit more.

> ---
>  tests/avocado/boot_linux.py | 6 +++++-
>  1 file changed, 5 insertions(+), 1 deletion(-)
>
> diff --git a/tests/avocado/boot_linux.py b/tests/avocado/boot_linux.py
> index be30dcbd58..c3869a987c 100644
> --- a/tests/avocado/boot_linux.py
> +++ b/tests/avocado/boot_linux.py
> @@ -91,9 +91,9 @@ class BootLinuxPPC64(LinuxTest):
>      :avocado: tags=arch:ppc64
>      """
>  
> +    # timeout for downloading new VM image.
>      timeout = 360
>  
> -    @skipIf(os.getenv('GITLAB_CI'), 'Running on GitLab')
>      def test_pseries_tcg(self):
>          """
>          :avocado: tags=machine:pseries
> @@ -101,6 +101,10 @@ def test_pseries_tcg(self):
>          """
>          self.require_accelerator("tcg")
>          self.vm.add_args("-accel", "tcg")
> +
> +        # timeout for actual Linux PPC boot test
> +        self.timeout = 360
> +
>          self.launch_and_wait(set_up_ssh_connection=False)


-- 
Alex Bennée
Virtualisation Tech Lead @ Linaro