Off-by-one in the retries count in KilledWorker (#8203)

distributed/scheduler.py

@@ -8449,8 +8449,8 @@ def allowed_failures(self) -> int:
 
     def __str__(self) -> str:
         return (
-            f"Attempted to run task {self.task} on {self.allowed_failures} different "
-            "workers, but all those workers died while running it. "
+            f"Attempted to run task {self.task} on {self.allowed_failures + 1} "
+            "different workers, but all those workers died while running it. "
             f"The last worker that attempt to run the task was {self.last_worker.address}. "
             "Inspecting worker logs is often a good next step to diagnose what went wrong. "
             "For more information see https://distributed.dask.org/en/stable/killed.html."

distributed/tests/test_scheduler.py

@@ -4244,8 +4244,8 @@ async def test_KilledWorker_informative_message(s, a, b):
     with pytest.raises(KilledWorker) as excinfo:
         raise ex
     msg = str(excinfo.value)
-    assert "Attempted to run task foo-bar" in msg
-    assert str(s.allowed_failures) in msg
+    assert "Attempted to run task foo-bar on 667 different workers" in msg
+    assert a.address in msg
     assert "worker logs" in msg
     assert "https://distributed.dask.org/en/stable/killed.html" in msg