Better XDP error handling

[enhaut]

Description

XDP's test modules might raise an exception when running on agent
machine. Currently these are not propagated to conroller, so it treats
incomplete/wrong results as a valid ones and so, controller crashes
on weird issues because of that.

Tests

• no crash test (just to be sure everything works as expected) J:10515937
• crash test (to test added code) J:10515938

[olichtne]

So this is probably a good step but this just raises a different exception which means that the Controller will still crash, just with a different exception.

this can be seen in your crash test logs:

LNST Controller crashed with an exception:
Traceback (most recent call last):
  File "/mnt/tests/data.lnst.anl.eng.rdu2.dc.redhat.com/data-server-content/gitlab-tasks/beaker-lnst-tasks/master.tar.gz/lnst/test-runner/./do-my-test", line 35, in main
    ctl.run(recipe, multimatch=bool(params.get("MULTIMATCH", False)))
  File "/root/virtualenvs/rhextensions-lnst-Xo1BSm3a-py3.12/lib/python3.12/site-packages/lnst/Controller/Controller.py", line 172, in run
    recipe.test()
  File "/root/rhextensions-lnst/lnst/RHExtensions/RHRecipeMixin.py", line 109, in test
    super(RHRecipeMixin, self).test()
  File "/root/virtualenvs/rhextensions-lnst-Xo1BSm3a-py3.12/lib/python3.12/site-packages/lnst/Recipes/ENRT/BaseEnrtRecipe.py", line 210, in test
    self.do_tests(recipe_config)
  File "/root/virtualenvs/rhextensions-lnst-Xo1BSm3a-py3.12/lib/python3.12/site-packages/lnst/Recipes/ENRT/BaseEnrtRecipe.py", line 332, in do_tests
    self.do_perf_tests(recipe_config)
  File "/root/virtualenvs/rhextensions-lnst-Xo1BSm3a-py3.12/lib/python3.12/site-packages/lnst/Recipes/ENRT/BaseEnrtRecipe.py", line 357, in do_perf_tests
    result = self.perf_test(perf_config)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/rhextensions-lnst/lnst/RHExtensions/RHRecipeMixin.py", line 425, in perf_test
    return super().perf_test(recipe_conf)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/virtualenvs/rhextensions-lnst-Xo1BSm3a-py3.12/lib/python3.12/site-packages/lnst/RecipeCommon/Perf/Recipe.py", line 162, in perf_test
    self.perf_test_iteration(recipe_conf, results)
  File "/root/virtualenvs/rhextensions-lnst-Xo1BSm3a-py3.12/lib/python3.12/site-packages/lnst/RecipeCommon/Perf/Recipe.py", line 192, in perf_test_iteration
    measurement_results = measurement.collect_results()
                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/virtualenvs/rhextensions-lnst-Xo1BSm3a-py3.12/lib/python3.12/site-packages/lnst/RecipeCommon/Perf/Measurements/XDPBenchMeasurement.py", line 154, in collect_results
    flow_results.generator_results = self._parse_generator_results(
                                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/virtualenvs/rhextensions-lnst-Xo1BSm3a-py3.12/lib/python3.12/site-packages/lnst/RecipeCommon/Perf/Measurements/XDPBenchMeasurement.py", line 169, in _parse_generator_results
    raise job.result["exception"]  # propagate exception from agent
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
lnst.Tests.BaseTestModule.TestModuleError: pktgen module is not loaded

This is better than before, it better explains what went wrong, but our other Measurement modules don't reraise exceptions, instead they return "fail results" with some error messages to indicate issues.

This may be important when running multiple tests where possibly some can fail due to errors but some can run just fine.

Or potentially to let rerun mechanics to rerun the test automatically in case of random errors, similar to what i did here: #382

In some cases it's likely good to raise some exception, to indicate serious really bad issues, but i'm not sure if this should be always the case. In what situations will this occur?

[enhaut]

Hmm actually yeah, in context of our tooling it makes sense to just report "invalid" results