Avoid duplicate collection errors from workers

changelog/1140.bugfix.rst

@@ -0,0 +1 @@
+Deduplicate identical collection errors from multiple workers while preserving different errors and errors from different modules.

src/xdist/dsession.py

@@ -55,7 +55,7 @@ def __init__(self, config: pytest.Config) -> None:
         self.maxfail: int = config.getvalue("maxfail")
         self.queue: Queue[tuple[str, dict[str, Any]]] = Queue()
         self._session: pytest.Session | None = None
-        self._failed_collection_errors: dict[object, bool] = {}
+        self._failed_collection_errors: set[tuple[str, str]] = set()
         self._active_nodes: set[WorkerController] = set()
         self._failed_nodes_count = 0
         self._max_worker_restart = get_default_max_worker_restart(self.config)
@@ -410,8 +410,10 @@ def _failed_worker_collectreport(
     ) -> None:
         # Check we haven't already seen this report (from
         # another worker).
-        if rep.longrepr not in self._failed_collection_errors:
-            self._failed_collection_errors[rep.longrepr] = True
+        # Exception representations from different workers need not compare equal.
+        report_key = (rep.nodeid, str(rep.longrepr))
+        if report_key not in self._failed_collection_errors:
+            self._failed_collection_errors.add(report_key)
             self.config.hook.pytest_collectreport(report=rep)
             self._handlefailures(rep)
 

testing/acceptance_test.py

@@ -4,6 +4,7 @@
 import re
 import shutil
 from typing import cast
+import xml.etree.ElementTree as ET
 
 import pytest
 
@@ -1722,6 +1723,39 @@ def test_get_xdist_worker_id(self, fake_request: pytest.FixtureRequest) -> None:
         assert xdist.get_xdist_worker_id(fake_request) == "master"
 
 
+@pytest.mark.parametrize("error_source", ["same", "worker", "module"])
+def test_collection_error_deduplication(
+    pytester: pytest.Pytester, error_source: str
+) -> None:
+    source = "raise RuntimeError('collection failure')"
+    if error_source == "worker":
+        source = (
+            "import os\n"
+            "raise RuntimeError('failure from ' + os.environ['PYTEST_XDIST_WORKER'])"
+        )
+    pytester.makepyfile(test_bad=source)
+    if error_source == "module":
+        pytester.makepyfile(test_other=source)
+
+    result = pytester.runpytest("-n2", "--junitxml=report.xml")
+    errors = 1 if error_source == "same" else 2
+    result.assert_outcomes(errors=errors)
+    assert result.ret == pytest.ExitCode.TESTS_FAILED
+    suite = ET.parse(pytester.path / "report.xml").getroot().find("testsuite")
+    assert suite is not None
+    assert suite.get("errors") == str(errors)
+    reports = suite.findall(".//error")
+    assert len(reports) == errors
+    if error_source == "worker":
+        assert any("failure from gw0" in (report.text or "") for report in reports)
+        assert any("failure from gw1" in (report.text or "") for report in reports)
+    elif error_source == "module":
+        assert {case.get("name") for case in suite.findall("testcase")} == {
+            "test_bad",
+            "test_other",
+        }
+
+
 def test_collection_crash(pytester: pytest.Pytester) -> None:
     p1 = pytester.makepyfile(
         """