[PATCH 2/9] tests/avocado: mark boot_linux.py long runtime instead of flaky

Nicholas Piggin posted 9 patches 10 months, 3 weeks ago
Maintainers: "Alex Bennée" <alex.bennee@linaro.org>, "Philippe Mathieu-Daudé" <philmd@linaro.org>, Thomas Huth <thuth@redhat.com>, Wainer dos Santos Moschetta <wainersm@redhat.com>, Beraldo Leal <bleal@redhat.com>, Cleber Rosa <crosa@redhat.com>, Nicholas Piggin <npiggin@gmail.com>, Daniel Henrique Barboza <danielhb413@gmail.com>, "Cédric Le Goater" <clg@kaod.org>, David Gibson <david@gibson.dropbear.id.au>, Harsh Prateek Bora <harshpb@linux.ibm.com>, Pavel Dovgalyuk <pavel.dovgaluk@ispras.ru>, Paolo Bonzini <pbonzini@redhat.com>
There is a newer version of this series
[PATCH 2/9] tests/avocado: mark boot_linux.py long runtime instead of flaky
Posted by Nicholas Piggin 10 months, 3 weeks ago
The ppc64 and s390x tests were first marked skipIf GITLAB_CI by commit
c0c8687ef0f ("tests/avocado: disable BootLinuxPPC64 test in CI"), and
commit 0f26d94ec9e ("tests/acceptance: skip s390x_ccw_vrtio_tcg on
GitLab") due to being very heavy-weight for gitlab CI.

Commit 9b45cc99318 ("docs/devel: rationalise unstable gitlab tests under
FLAKY_TESTS") changed this to being flaky but it isn't really, it just
had a long runtime.

So introduce a new AVOCADO_ALLOW_LONG_RUNTIME variable and make these
tests require it. Re-testing the s390x and ppc64 tests on gitlab shows
about 100-150s runtime each, which is similar to the x86-64 tests.
Since these are among the longest running avocado tests, make x86-64
require long runtime as well.

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
---
 docs/devel/testing.rst      | 8 ++++++++
 tests/avocado/boot_linux.py | 8 ++------
 2 files changed, 10 insertions(+), 6 deletions(-)

diff --git a/docs/devel/testing.rst b/docs/devel/testing.rst
index bd132306c1..3a9c1327be 100644
--- a/docs/devel/testing.rst
+++ b/docs/devel/testing.rst
@@ -1346,6 +1346,14 @@ the environment.
 The definition of *large* is a bit arbitrary here, but it usually means an
 asset which occupies at least 1GB of size on disk when uncompressed.
 
+AVOCADO_ALLOW_LONG_RUNTIME
+^^^^^^^^^^^^^^^^^^^^^^^^^^
+Tests which have a long runtime will not be run unless that
+``AVOCADO_ALLOW_LONG_RUNTIME=1`` is exported on the environment.
+
+The definition of *long* is a bit arbitrary here, but it usually means a
+test which takes more than 100 seconds to complete.
+
 AVOCADO_ALLOW_UNTRUSTED_CODE
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 There are tests which will boot a kernel image or firmware that can be
diff --git a/tests/avocado/boot_linux.py b/tests/avocado/boot_linux.py
index 7c4769904e..6df0fc0489 100644
--- a/tests/avocado/boot_linux.py
+++ b/tests/avocado/boot_linux.py
@@ -93,13 +93,11 @@ class BootLinuxPPC64(LinuxTest):
 
     timeout = 360
 
-    @skipUnless(os.getenv('QEMU_TEST_FLAKY_TESTS'), 'Test is unstable on GitLab')
-
+    @skipUnless(os.getenv('AVOCADO_ALLOW_LONG_RUNTIME'), 'runtime limited')
     def test_pseries_tcg(self):
         """
         :avocado: tags=machine:pseries
         :avocado: tags=accel:tcg
-        :avocado: tags=flaky
         """
         self.require_accelerator("tcg")
         self.vm.add_args("-accel", "tcg")
@@ -113,13 +111,11 @@ class BootLinuxS390X(LinuxTest):
 
     timeout = 240
 
-    @skipUnless(os.getenv('QEMU_TEST_FLAKY_TESTS'), 'Test is unstable on GitLab')
-
+    @skipUnless(os.getenv('AVOCADO_ALLOW_LONG_RUNTIME'), 'runtime limited')
     def test_s390_ccw_virtio_tcg(self):
         """
         :avocado: tags=machine:s390-ccw-virtio
         :avocado: tags=accel:tcg
-        :avocado: tags=flaky
         """
         self.require_accelerator("tcg")
         self.vm.add_args("-accel", "tcg")
-- 
2.42.0
Re: [PATCH 2/9] tests/avocado: mark boot_linux.py long runtime instead of flaky
Posted by Thomas Huth 10 months, 3 weeks ago
On 07/01/2024 18.01, Nicholas Piggin wrote:
> The ppc64 and s390x tests were first marked skipIf GITLAB_CI by commit
> c0c8687ef0f ("tests/avocado: disable BootLinuxPPC64 test in CI"), and
> commit 0f26d94ec9e ("tests/acceptance: skip s390x_ccw_vrtio_tcg on
> GitLab") due to being very heavy-weight for gitlab CI.
> 
> Commit 9b45cc99318 ("docs/devel: rationalise unstable gitlab tests under
> FLAKY_TESTS") changed this to being flaky but it isn't really, it just
> had a long runtime.
> 
> So introduce a new AVOCADO_ALLOW_LONG_RUNTIME variable and make these
> tests require it. Re-testing the s390x and ppc64 tests on gitlab shows
> about 100-150s runtime each, which is similar to the x86-64 tests.
> Since these are among the longest running avocado tests, make x86-64
> require long runtime as well.
> 
> Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
> ---
>   docs/devel/testing.rst      | 8 ++++++++
>   tests/avocado/boot_linux.py | 8 ++------
>   2 files changed, 10 insertions(+), 6 deletions(-)
> 
> diff --git a/docs/devel/testing.rst b/docs/devel/testing.rst
> index bd132306c1..3a9c1327be 100644
> --- a/docs/devel/testing.rst
> +++ b/docs/devel/testing.rst
> @@ -1346,6 +1346,14 @@ the environment.
>   The definition of *large* is a bit arbitrary here, but it usually means an
>   asset which occupies at least 1GB of size on disk when uncompressed.
>   
> +AVOCADO_ALLOW_LONG_RUNTIME
> +^^^^^^^^^^^^^^^^^^^^^^^^^^
> +Tests which have a long runtime will not be run unless that
> +``AVOCADO_ALLOW_LONG_RUNTIME=1`` is exported on the environment.
> +
> +The definition of *long* is a bit arbitrary here, but it usually means a
> +test which takes more than 100 seconds to complete.

For the qtests (and others), we are using "make check SPEED=slow", so what 
about just re-using the SPEED environment variable instead? IMHO that would 
be more consistent?

  Thomas
Re: [PATCH 2/9] tests/avocado: mark boot_linux.py long runtime instead of flaky
Posted by Cédric Le Goater 10 months, 3 weeks ago
On 1/7/24 18:01, Nicholas Piggin wrote:
> The ppc64 and s390x tests were first marked skipIf GITLAB_CI by commit
> c0c8687ef0f ("tests/avocado: disable BootLinuxPPC64 test in CI"), and
> commit 0f26d94ec9e ("tests/acceptance: skip s390x_ccw_vrtio_tcg on
> GitLab") due to being very heavy-weight for gitlab CI.
> 
> Commit 9b45cc99318 ("docs/devel: rationalise unstable gitlab tests under
> FLAKY_TESTS") changed this to being flaky but it isn't really, it just
> had a long runtime.

I agree. The same comment applies to the aspeed tests.
  
> So introduce a new AVOCADO_ALLOW_LONG_RUNTIME variable and make these
> tests require it. Re-testing the s390x and ppc64 tests on gitlab shows
> about 100-150s runtime each, which is similar to the x86-64 tests.
> Since these are among the longest running avocado tests, make x86-64
> require long runtime as well.
> 
> Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
>
> ---
>   docs/devel/testing.rst      | 8 ++++++++
>   tests/avocado/boot_linux.py | 8 ++------
>   2 files changed, 10 insertions(+), 6 deletions(-)
> 
> diff --git a/docs/devel/testing.rst b/docs/devel/testing.rst
> index bd132306c1..3a9c1327be 100644
> --- a/docs/devel/testing.rst
> +++ b/docs/devel/testing.rst
> @@ -1346,6 +1346,14 @@ the environment.
>   The definition of *large* is a bit arbitrary here, but it usually means an
>   asset which occupies at least 1GB of size on disk when uncompressed.
>   
> +AVOCADO_ALLOW_LONG_RUNTIME
> +^^^^^^^^^^^^^^^^^^^^^^^^^^
> +Tests which have a long runtime will not be run unless that
> +``AVOCADO_ALLOW_LONG_RUNTIME=1`` is exported on the environment.
> +
> +The definition of *long* is a bit arbitrary here, but it usually means a
> +test which takes more than 100 seconds to complete.
Defining a limit is the difficult part. The Aspeed SDK tests have been
tagged flaky for CI even though they run in less than 50s. Anyhow,

Reviewed-by: Cédric Le Goater <clg@kaod.org>

Thanks,

C.




> +
>   AVOCADO_ALLOW_UNTRUSTED_CODE
>   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>   There are tests which will boot a kernel image or firmware that can be
> diff --git a/tests/avocado/boot_linux.py b/tests/avocado/boot_linux.py
> index 7c4769904e..6df0fc0489 100644
> --- a/tests/avocado/boot_linux.py
> +++ b/tests/avocado/boot_linux.py
> @@ -93,13 +93,11 @@ class BootLinuxPPC64(LinuxTest):
>   
>       timeout = 360
>   
> -    @skipUnless(os.getenv('QEMU_TEST_FLAKY_TESTS'), 'Test is unstable on GitLab')
> -
> +    @skipUnless(os.getenv('AVOCADO_ALLOW_LONG_RUNTIME'), 'runtime limited')
>       def test_pseries_tcg(self):
>           """
>           :avocado: tags=machine:pseries
>           :avocado: tags=accel:tcg
> -        :avocado: tags=flaky
>           """
>           self.require_accelerator("tcg")
>           self.vm.add_args("-accel", "tcg")
> @@ -113,13 +111,11 @@ class BootLinuxS390X(LinuxTest):
>   
>       timeout = 240
>   
> -    @skipUnless(os.getenv('QEMU_TEST_FLAKY_TESTS'), 'Test is unstable on GitLab')
> -
> +    @skipUnless(os.getenv('AVOCADO_ALLOW_LONG_RUNTIME'), 'runtime limited')
>       def test_s390_ccw_virtio_tcg(self):
>           """
>           :avocado: tags=machine:s390-ccw-virtio
>           :avocado: tags=accel:tcg
> -        :avocado: tags=flaky
>           """
>           self.require_accelerator("tcg")
>           self.vm.add_args("-accel", "tcg")