fix: reconcile live scheduled resume cancellation

tests/wf_server/test_scheduler_integration.py

@@ -1408,6 +1408,87 @@ async def test_shutdown_timeout_fences_inflight_resume_and_restart_keeps_decisio
         await service.stop()
 
 
+async def test_caller_cancellation_reconciles_and_frees_capacity(
+    tmp_path: Path,
+) -> None:
+    """Caller cancellation fails one ambiguous resume without fencing capacity."""
+    _gate_open.clear()
+    root = tmp_path / "store"
+    server = build_local_static_workflow_server(root, extra_sources=_gate_sources())
+    await _seed(
+        server,
+        "ask_gate",
+        "ask_gate.default",
+        _interrupt_then_gate_plan("ask_gate"),
+        ["submitted"],
+    )
+    await _seed(server, "const", "const.default", _constant_plan("const"), ["ok"])
+    store = FileScheduleStore(root)
+    store.create_schedule(_one_shot("asker", "ask_gate.default", datetime.now(UTC)))
+    service = _scheduler(server, capacity=1, auto_tick=False)
+    try:
+        await service.start()
+        async with asyncio.timeout(20):
+            while not _run_ids(root):
+                await service.poll_once(datetime.now(UTC))
+                await asyncio.sleep(0.02)
+        ask_id = _run_ids(root)[0]
+        await _wait_for(
+            lambda: FileRunStore(root).get_run(ask_id).status.value == "interrupted"
+        )
+
+        resume_task = asyncio.create_task(
+            server.api.resume_run(
+                run_id=ask_id, resume_payload={}, resume_outcome="submitted"
+            )
+        )
+        await _wait_for(lambda: ask_id in service._live_resumes)
+        await _wait_for(lambda: FileRunStore(root).is_executing(ask_id))
+        await _wait_for(
+            lambda: (
+                (attempt := FileRunStore(root).get_resume_attempt(ask_id)) is not None
+                and attempt.state == "ACTIVE"
+            )
+        )
+        resume_task.cancel()
+        with pytest.raises(asyncio.CancelledError):
+            await resume_task
+
+        assert service.running
+        assert not service._live_resumes
+        cancelled = FileRunStore(root).get_run(ask_id)
+        assert cancelled.status.value == "failed"
+        assert not FileRunStore(root).is_executing(ask_id)
+        attempt = FileRunStore(root).get_resume_attempt(ask_id)
+        assert attempt is not None and attempt.state == "ACTIVE"
+        assert len(_entries(root, "asker", "failed")) == 1
+
+        with pytest.raises(ValueError, match="ambiguous active resume attempt"):
+            await server.api.resume_run(
+                run_id=ask_id, resume_payload={}, resume_outcome="submitted"
+            )
+        assert _entries(root, "asker", "completed") == []
+
+        store.create_schedule(_one_shot("after", "const.default", datetime.now(UTC)))
+        async with asyncio.timeout(20):
+            while True:
+                await service.poll_once(datetime.now(UTC))
+                if len(_run_ids(root)) == 2:
+                    after_id = [rid for rid in _run_ids(root) if rid != ask_id][0]
+                    if FileRunStore(root).get_run(after_id).status.value == "completed":
+                        break
+                await asyncio.sleep(0.02)
+        assert await _kinds(root, "asker") == [
+            "admitted",
+            "interrupted",
+            "failed",
+        ]
+        assert await _kinds(root, "after") == ["admitted", "completed"]
+    finally:
+        _gate_open.set()
+        await service.stop()
+
+
 async def test_scheduled_resume_requested_during_drain_rejected(
     tmp_path: Path,
 ) -> None: