feat(swarm): expose needs_resync on TopologySummary + upsert record_error

Two small observability follow-ups to the phase-1 agent/topology wiring:

TopologySummary now carries needs_resync so operators can see the
heartbeat's resync flag via the topology list/detail API without
dropping into the DB.

TopologyStore.record_error becomes an upsert — when a docker/compose
failure fires during the first materialise (put() never reached), we
still land a marker row so GET /topology/state surfaces the error and
the next heartbeat carries an empty applied_version_hash. That empty
hash is what master's heartbeat check relies on to flag the topology
for resync instead of assuming the apply succeeded.

decnet/agent/topology_store.py

@@ -131,10 +131,22 @@ class TopologyStore:
         self._conn.commit()
 
     def record_error(self, topology_id: str, message: str) -> None:
-        """Attach a last-error message to the current row (for debugging)."""
+        """Attach a last-error message for *topology_id*.
+
+        Upserts a marker row when no apply has yet succeeded for this
+        topology — that way a failure *during* the first materialise
+        (put() hasn't been reached) still surfaces via GET
+        /topology/state and the next heartbeat.  The marker row uses an
+        empty ``applied_version_hash`` so master's heartbeat check sees
+        the hash mismatch and schedules a resync.
+        """
         self._conn.execute(
-            "UPDATE applied_topology SET last_error=? WHERE topology_id=?",
-            (message, topology_id),
+            "INSERT INTO applied_topology"
+            " (topology_id, applied_version_hash, hydrated_blob_json,"
+            "  applied_at, last_error)"
+            " VALUES (?, '', '{}', 0, ?)"
+            " ON CONFLICT(topology_id) DO UPDATE SET last_error=excluded.last_error",
+            (topology_id, message),
         )
         self._conn.commit()
 

decnet/web/db/models.py

@@ -687,6 +687,7 @@ class TopologySummary(BaseModel):
     target_host_uuid: Optional[str] = None
     status: str
     version: int
+    needs_resync: bool = False
     created_at: datetime
     status_changed_at: Optional[datetime] = None
 

tests/api/topology/test_models.py

@@ -84,6 +84,22 @@ async def test_summary_accepts_repo_topology_row(repo):
     summary = TopologySummary(**row)
     assert summary.id == tid
     assert summary.version == 1
+    # Defaults surface cleanly on a fresh topology.
+    assert summary.needs_resync is False
+    assert summary.target_host_uuid is None
+
+
+@pytest.mark.anyio
+async def test_summary_surfaces_needs_resync_flag(repo):
+    """When the heartbeat handler flags a topology for resync, the API
+    list/detail views must expose it so operators can debug without
+    shelling into the DB."""
+    plan = generate(_cfg())
+    tid = await persist(repo, plan)
+    await repo.set_topology_resync(tid, True)
+    row = await repo.get_topology(tid)
+    summary = TopologySummary(**row)
+    assert summary.needs_resync is True
 
 
 @pytest.mark.anyio

tests/swarm/test_agent_topology_endpoints.py

@@ -101,6 +101,21 @@ def test_topology_apply_docker_failure_is_500_and_records_error(
         raise RuntimeError("docker down")
 
     monkeypatch.setattr(_ops, "apply", _boom)
+
+    # Stub docker.from_env for the /topology/state observed() call so
+    # the state endpoint doesn't need a real daemon.
+    class _StubDocker:
+        class networks:
+            @staticmethod
+            def list(): return []
+
+        class containers:
+            @staticmethod
+            def list(all=False): return []
+
+    import docker as _docker
+    monkeypatch.setattr(_docker, "from_env", lambda: _StubDocker)
+
     client = TestClient(_agent_app.app)
     resp = client.post(
         "/topology/apply",
@@ -109,6 +124,14 @@ def test_topology_apply_docker_failure_is_500_and_records_error(
     assert resp.status_code == 500
     assert "docker down" in resp.json()["detail"]
 
+    # The error must be persisted so GET /topology/state surfaces it,
+    # and the stored hash stays empty so master's heartbeat check flags
+    # the topology for resync rather than assuming it's applied.
+    state = client.get("/topology/state").json()
+    assert state["topology_id"] == "top-err"
+    assert state["applied_version_hash"] == ""
+    assert state["last_error"] == "docker down"
+
 
 def test_topology_teardown_routes_to_ops(monkeypatch: pytest.MonkeyPatch) -> None:
     called: dict = {}

tests/swarm/test_agent_topology_store.py

@@ -64,6 +64,34 @@ def test_record_error_then_put_clears(tmp_path: pathlib.Path) -> None:
     s.close()
 
 
+def test_record_error_upserts_when_no_prior_row(tmp_path: pathlib.Path) -> None:
+    """Apply failure mid-materialise: put() hasn't written a row yet but
+    we still want the error surfaced on GET /topology/state and the
+    next heartbeat.  The marker uses empty hash so master sees drift."""
+    s = _store(tmp_path)
+    s.record_error("t-fail", "docker refused connection")
+    row = s.current()
+    assert row is not None
+    assert row.topology_id == "t-fail"
+    assert row.applied_version_hash == ""
+    assert row.applied_at == 0
+    assert row.last_error == "docker refused connection"
+    s.close()
+
+
+def test_record_error_then_successful_put_replaces_marker(tmp_path: pathlib.Path) -> None:
+    """Once a retry succeeds, the marker row must be replaced with a
+    real applied row — no stale error or empty hash left behind."""
+    s = _store(tmp_path)
+    s.record_error("t-retry", "first try failed")
+    s.put("t-retry", "real-hash", {"topology": {"id": "t-retry"}})
+    row = s.current()
+    assert row.applied_version_hash == "real-hash"
+    assert row.last_error is None
+    assert row.applied_at > 0
+    s.close()
+
+
 def test_clear(tmp_path: pathlib.Path) -> None:
     s = _store(tmp_path)
     s.put("t1", "h", {})