Extract health-check crash-detection logic into runtime state helper

backend/tests/test_tasks/test_health_check.py

@@ -77,7 +77,7 @@ class TestRunProbe:
             "app.tasks.health_check.health_service.probe",
             new_callable=AsyncMock,
             return_value=new_status,
-        ), patch("app.tasks.health_check.log") as mock_log:
+        ), patch("app.utils.runtime_state.log") as mock_log:
             await _run_probe(app)
 
         online_calls = [c for c in mock_log.info.call_args_list if c[0][0] == "fail2ban_came_online"]
@@ -93,7 +93,7 @@ class TestRunProbe:
             "app.tasks.health_check.health_service.probe",
             new_callable=AsyncMock,
             return_value=new_status,
-        ), patch("app.tasks.health_check.log") as mock_log:
+        ), patch("app.utils.runtime_state.log") as mock_log:
             await _run_probe(app)
 
         offline_calls = [c for c in mock_log.warning.call_args_list if c[0][0] == "fail2ban_went_offline"]
@@ -109,7 +109,7 @@ class TestRunProbe:
             "app.tasks.health_check.health_service.probe",
             new_callable=AsyncMock,
             return_value=new_status,
-        ), patch("app.tasks.health_check.log") as mock_log:
+        ), patch("app.utils.runtime_state.log") as mock_log:
             await _run_probe(app)
 
         transition_calls = [
@@ -134,7 +134,7 @@ class TestRunProbe:
             "app.tasks.health_check.health_service.probe",
             new_callable=AsyncMock,
             return_value=new_status,
-        ), patch("app.tasks.health_check.log") as mock_log:
+        ), patch("app.utils.runtime_state.log") as mock_log:
             await _run_probe(app)
 
         transition_calls = [
@@ -180,7 +180,7 @@ class TestRunProbe:
                 new_callable=AsyncMock,
                 return_value=new_status,
             ),
-            patch("app.tasks.health_check.log"),
+            patch("app.utils.runtime_state.log"),
         ):
             # Must not raise even with no prior status.
             await _run_probe(app)

backend/tests/test_utils/test_runtime_state.py

@@ -1,8 +1,11 @@
 from __future__ import annotations
 
+import datetime
 from unittest.mock import MagicMock
 
 from app.config import Settings
+from app.models.config import PendingRecovery
+from app.models.server import ServerStatus
 from app.utils.runtime_state import get_app_settings, get_effective_settings
 
 
@@ -45,3 +48,40 @@ def test_get_app_settings_reads_bootstrap_settings() -> None:
     app = _FakeApp(_FakeState(settings=settings))
 
     assert get_app_settings(app) is settings
+
+
+def test_process_health_probe_result_creates_pending_recovery_within_window() -> None:
+    from app.utils.runtime_state import RuntimeState, process_health_probe_result
+
+    now = datetime.datetime.now(tz=datetime.UTC)
+    runtime_state = RuntimeState(
+        server_status=ServerStatus(online=True),
+        last_activation={"jail_name": "sshd", "at": now - datetime.timedelta(seconds=30)},
+        pending_recovery=None,
+    )
+
+    process_health_probe_result(runtime_state, ServerStatus(online=False), now=now)
+
+    assert runtime_state.pending_recovery is not None
+    assert runtime_state.pending_recovery.jail_name == "sshd"
+    assert runtime_state.pending_recovery.recovered is False
+
+
+def test_process_health_probe_result_resolves_existing_pending_recovery() -> None:
+    from app.utils.runtime_state import RuntimeState, process_health_probe_result
+
+    activated_at = datetime.datetime.now(tz=datetime.UTC) - datetime.timedelta(seconds=30)
+    runtime_state = RuntimeState(
+        server_status=ServerStatus(online=False),
+        pending_recovery=PendingRecovery(
+            jail_name="sshd",
+            activated_at=activated_at,
+            detected_at=activated_at + datetime.timedelta(seconds=10),
+            recovered=False,
+        ),
+    )
+
+    process_health_probe_result(runtime_state, ServerStatus(online=True), now=activated_at + datetime.timedelta(seconds=20))
+
+    assert runtime_state.pending_recovery is not None
+    assert runtime_state.pending_recovery.recovered is True