fix(edge): resilient multi-camera startup — skip a camera that fails to construct

build_runtimes() builds each camera's source+runtime in a try/except: a camera that
raises at construction (e.g. an unseated/absent CSI camera whose Picamera2(...) throws)
is logged and skipped so the working cameras still run, instead of an unguarded
exception crash-looping the whole daemon (Restart=always). Raises only if no camera
comes up. Closes the eager-construction gap the review flagged.

edge/daemon.py

@@ -65,6 +65,25 @@ def parse_cameras(env, default_camera_id):
     return [(backend, device, default_camera_id)]
 
 
+def build_runtimes(cameras, make_source, make_runtime, log=None):
+    """Build one (source -> runtime) per camera. A camera that fails to CONSTRUCT
+    (e.g. an unseated/absent CSI camera whose Picamera2(...) raises) is logged and
+    SKIPPED so it never crash-loops the whole daemon and the other cameras still run.
+    Raises if no camera comes up (never run monitoring nothing)."""
+    if log is None:
+        import sys
+        log = lambda m: print(m, file=sys.stderr)
+    runtimes = []
+    for backend, device, cam in cameras:
+        try:
+            runtimes.append(make_runtime(cam, make_source(backend, device)))
+        except Exception as e:
+            log(f"camera {cam} ({backend}:{device}) failed to start, skipping: {e}")
+    if not runtimes:
+        raise RuntimeError("no cameras could be started")
+    return runtimes
+
+
 def main() -> None:  # pragma: no cover (integration entrypoint, exercised on device)
     import time
     from dataclasses import replace
@@ -101,11 +120,15 @@ def main() -> None:  # pragma: no cover (integration entrypoint, exercised on de
     queue = OfflineQueue(os.environ.get("QUEUE_DB", "queue.db"), post)
 
     pose = build_pose()  # shared across cameras
-    runtimes = []
-    for backend, device, cam in parse_cameras(dict(os.environ), cfg.camera_id):
+
+    def make_runtime(cam, source):
         # Per-camera config so events are stamped with that camera's id.
         cam_cfg = replace(cfg, camera_id=cam, engine=replace(cfg.engine, camera_id=cam))
-        runtimes.append(PipelineRuntime(cam_cfg, build_source(backend, device), pose, queue))
+        return PipelineRuntime(cam_cfg, source, pose, queue)
+
+    # A missing/unseated camera is skipped (logged), not fatal — the others still run.
+    runtimes = build_runtimes(parse_cameras(dict(os.environ), cfg.camera_id),
+                              build_source, make_runtime)
 
     run_loop(runtimes, client, queue, clock=now, sleep=time.sleep, iterations=None,
              frame_interval=1.0 / float(os.environ.get("TARGET_FPS", "12")),

tests/test_multicam.py

@@ -1,7 +1,9 @@
 """TDD for multi-camera support: per-camera zone routing, a run_loop that steps
 N cameras per tick, and CAMERAS env parsing. All with fakes (no hardware)."""
+import pytest
+
 from edge.zones import zones_from_config
-from edge.daemon import run_loop, parse_cameras
+from edge.daemon import run_loop, parse_cameras, build_runtimes
 
 
 def _cfg_two_cams():
@@ -73,10 +75,30 @@ def test_parse_cameras_multi_and_single():
 
 
 def test_parse_cameras_rejects_blank_entries_never_zero_cameras():
-    import pytest
     # A non-empty-but-all-blank CAMERAS must fail loudly, never yield zero cameras
     # (a safety daemon monitoring nothing while looking healthy).
     with pytest.raises(ValueError):
         parse_cameras({"CAMERAS": " , "}, "cam-01")
     with pytest.raises(ValueError):
         parse_cameras({"CAMERAS": "csi:0"}, "cam-01")   # wrong arity
+
+
+def test_build_runtimes_skips_a_camera_that_fails_to_start():
+    # An unseated/absent CSI camera raises at construction; it must be skipped so the
+    # working USB camera still runs (construction resilience, not just step isolation).
+    def make_source(backend, device):
+        if backend == "csi":
+            raise RuntimeError("no CSI camera attached")
+        return ("src", backend, device)
+    logs = []
+    rts = build_runtimes([("csi", "0", "cam-01"), ("uvc", "0", "cam-02")],
+                         make_source, lambda cam, src: (cam, src), log=logs.append)
+    assert rts == [("cam-02", ("src", "uvc", "0"))]
+    assert any("cam-01" in m for m in logs)
+
+
+def test_build_runtimes_raises_when_all_cameras_fail():
+    def boom(*_a):
+        raise RuntimeError("gone")
+    with pytest.raises(RuntimeError):
+        build_runtimes([("csi", "0", "c")], boom, lambda c, s: None, log=lambda m: None)