[bug] samples/core/exit_handler/exit_handler_test.py is failing

[hbelmiro]

samples/core/exit_handler/exit_handler_test.py is failing

Full logs:

I0610 14:24:34.625197      26 launcher_v2.go:90] input ComponentSpec:{
  "executorLabel": "exec-fail-op",
  "inputDefinitions": {
    "parameters": {
      "message": {
        "parameterType": "STRING"
      }
    }
  }
}
I0610 14:24:34.625952      26 cache.go:116] Connecting to cache endpoint 10.96.159.44:8887
WARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv
[KFP Executor 2024-06-10 14:24:40,767 INFO]: Looking for component `fail_op` in --component_module_path `/tmp/tmp.fmftAEDVvr/ephemeral_component.py`
[KFP Executor 2024-06-10 14:24:40,767 INFO]: Loading KFP component "fail_op" from /tmp/tmp.fmftAEDVvr/ephemeral_component.py (directory "/tmp/tmp.fmftAEDVvr" and module name "ephemeral_component")
[KFP Executor 2024-06-10 14:24:40,768 INFO]: Got executor_input:
{
    "inputs": {
        "parameterValues": {
            "message": "Task failed."
        }
    },
    "outputs": {
        "outputFile": "/tmp/kfp_outputs/output_metadata.json"
    }
}
Task failed.
/usr/local/lib/python3.7/runpy.py:109: FutureWarning: Python 3.7 has reached end-of-life. KFP will drop support for Python 3.7 on April 23, 2024. To use new versions of the KFP SDK after that date, you will need to upgrade to Python >= 3.8. See https://devguide.python.org/versions/ for more details.
  __import__(pkg_name)
I0610 14:24:40.791065      26 launcher_v2.go:151] publish success.
F0610 14:24:40.791111      26 main.go:49] failed to execute component: exit status 1

Steps to reproduce

• Enable the tests in:

  • pipelines/.github/workflows/e2e-test.yaml

    Line 72 in 4dab191

    # Disabled while https://github.com/kubeflow/pipelines/issues/10885 is not resolved

• Run the CI

Reproducing locally

• Deploy KFP to a cluster.

• Forward the API and UI ports:

  kubectl port-forward -n kubeflow svc/ml-pipeline 8888:8888

  kubectl port-forward -n kubeflow svc/ml-pipeline-ui 8080:80

• Tweak the test to use your cluster

  • In

    pipelines/test/sample-test/constants.py

    Line 22 in f328f0b

    BASE_DIR = os.path.join('/python/src/github.com/', GITHUB_REPO)

    , use BASE_DIR = os.path.abspath('./')
  • Remove:

    • pipelines/test/sample-test/sample_test_launcher.py

      Line 59 in f328f0b

      self._bucket_name = results_gcs_dir.split('/')[2]

    • pipelines/test/sample-test/sample_test_launcher.py

      Line 73 in f328f0b

      inverse_proxy_config = v1.read_namespaced_config_map(

  • Replace

    pipelines/test/sample-test/sample_test_launcher.py

    Line 75 in f328f0b

    self._host = inverse_proxy_config.data.get('Hostname')

    with self._host = 'http://localhost:8888'

• Run:

  python ./test/sample-test/sample_test_launcher.py sample_test run_test --namespace kubeflow --test-name exit_handler --results-gcs-dir output

  The command should fail with exit code 1.

Labels

/area testing

---

Impacted by this bug? Give it a 👍.

[hbelmiro]

May be related to #10917.

[liavweiss]

/assign

[hbelmiro]

Resolved by #11100.