Keep the job thread alive after an error of the queue

An error outside a job, such as a full disk, stopped the thread, and no
job ran until a restart. The health check now answers 503 when a job
thread stopped.

tests/test_worker_flows.py

@@ -6,7 +6,7 @@ from unittest.mock import Mock
 
 import pytest
 
-from bindery import config, fff
+from bindery import config, fff, web
 from bindery import orchestrator as o
 from bindery.diffing import text_hash
 from bindery.jobs import DownloadMode, Job, JobQueue, QueueStatus
@@ -87,6 +87,42 @@ def test_worker_persists_outcomes_and_continues_after_failure(state, monkeypatch
     assert [c.args[0] for c in sleeps.call_args_list] == [2, 7]
 
 
+def test_worker_goes_on_after_an_error_of_the_queue(state, monkeypatch, caplog):
+    store, queue = state
+    job = queue.add(Job(mode=DownloadMode.REBUILD))
+    monkeypatch.setattr(queue, "get_next", Mock(side_effect=[OSError("disk full"), job, StopLoop()]))
+    monkeypatch.setattr(queue, "update", Mock(side_effect=[OSError("disk full"), None, None]))
+    sleeps = Mock()
+    monkeypatch.setattr(o.time, "sleep", sleeps)
+    with pytest.raises(StopLoop):
+        o.download_loop()
+    assert caplog.text.count("Job loop error: disk full") == 2
+    assert [c.args[0] for c in sleeps.call_args_list] == [2, 2]
+
+
+def test_health_reports_a_stopped_job_thread(state, monkeypatch):
+    client = web.app.test_client()
+    assert o.stopped_threads() == []
+    alive, dead = Mock(), Mock()
+    alive.is_alive.return_value, dead.is_alive.return_value, dead.name = True, False, "download_loop"
+    monkeypatch.setattr(o, "_threads", [alive, dead])
+    response = client.get("/api/health")
+    assert response.status_code == 503
+    assert response.json == {"status": "error", "error": "Stopped: download_loop. Restart Bindery."}
+    monkeypatch.setattr(o, "_threads", [alive])
+    assert client.get("/api/health").json == {"status": "ok"}
+
+
+def test_start_threads_keeps_the_threads_for_the_health_check(state, monkeypatch):
+    monkeypatch.setattr(o, "_threads", [])
+    monkeypatch.setattr(o, "download_loop", lambda: None)
+    monkeypatch.setattr(o, "scheduler_loop", lambda: None)
+    o.start_threads()
+    for thread in o._threads:
+        thread.join()
+    assert o.stopped_threads() == ["<lambda>", "<lambda>"]
+
+
 def test_scheduler_skips_disabled_busy_and_recent_stories(state, monkeypatch):
     store, queue = state
     now = datetime(2026, 6, 1, 12)