[CI] Automatic rerun of certain test failures (#36694)

This commit is contained in:
Joao Gante
2025-03-13 15:40:23 +00:00
committed by GitHub
parent d84569387f
commit a3201cea14
7 changed files with 44 additions and 11 deletions

View File

@@ -33,6 +33,26 @@ COMMON_ENV_VARIABLES = {
COMMON_PYTEST_OPTIONS = {"max-worker-restart": 0, "dist": "loadfile", "vvv": None, "rsfE":None}
DEFAULT_DOCKER_IMAGE = [{"image": "cimg/python:3.8.12"}]
# Strings that commonly appear in the output of flaky tests when they fail. These are used with `pytest-rerunfailures`
# to rerun the tests that match these patterns.
FLAKY_TEST_FAILURE_PATTERNS = [
"OSError", # Machine/connection transient error
"Timeout", # Machine/connection transient error
"ConnectionError", # Connection transient error
"FileNotFoundError", # Raised by `datasets` on Hub failures
"PIL.UnidentifiedImageError", # Raised by `PIL.Image.open` on connection issues
"HTTPError.*502", # Hub-related
"HTTPError.*504", # Hub-related
"AssertionError: Tensor-likes are not close!", # `torch.testing.assert_close`, we might have unlucky random values
# TODO: error downloading tokenizer's `merged.txt` from hub can cause all the exceptions below. Throw and handle
# them under a single message.
"TypeError: expected str, bytes or os.PathLike object, not NoneType",
"TypeError: stat: path should be string, bytes, os.PathLike or integer, not NoneType",
"Converting from Tiktoken failed",
"KeyError: <class ",
"TypeError: not a string",
]
class EmptyJob:
job_name = "empty"
@@ -124,7 +144,9 @@ class CircleCIJob:
# Examples special case: we need to download NLTK files in advance to avoid cuncurrency issues
timeout_cmd = f"timeout {self.command_timeout} " if self.command_timeout else ""
marker_cmd = f"-m '{self.marker}'" if self.marker is not None else ""
additional_flags = f" -p no:warning -o junit_family=xunit1 --junitxml=test-results/junit.xml"
junit_flags = f" -p no:warning -o junit_family=xunit1 --junitxml=test-results/junit.xml"
joined_flaky_patterns = "|".join(FLAKY_TEST_FAILURE_PATTERNS)
repeat_on_failure_flags = f"--reruns 5 --reruns-delay 2 --only-rerun '({joined_flaky_patterns})'"
parallel = f' << pipeline.parameters.{self.job_name}_parallelism >> '
steps = [
"checkout",
@@ -152,7 +174,7 @@ class CircleCIJob:
},
{"run": {
"name": "Run tests",
"command": f"({timeout_cmd} python3 -m pytest {marker_cmd} -n {self.pytest_num_workers} {additional_flags} {' '.join(pytest_flags)} $(cat splitted_tests.txt) | tee tests_output.txt)"}
"command": f"({timeout_cmd} python3 -m pytest {marker_cmd} -n {self.pytest_num_workers} {junit_flags} {repeat_on_failure_flags} {' '.join(pytest_flags)} $(cat splitted_tests.txt) | tee tests_output.txt)"}
},
{"run": {"name": "Expand to show skipped tests", "when": "always", "command": f"python3 .circleci/parse_test_outputs.py --file tests_output.txt --skip"}},
{"run": {"name": "Failed tests: show reasons", "when": "always", "command": f"python3 .circleci/parse_test_outputs.py --file tests_output.txt --fail"}},