From 581f337021a1d2c9f7e66854e91fe187e69ba0f4 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=EC=9C=A0=EC=B0=BD=EC=9A=B1?= <changukyu@gmail.com>
Date: Sun, 12 Jul 2026 11:07:27 +0900
Subject: [PATCH] fix(edge): daemon periodic ops (sync/heartbeat/flush) can't
 crash the loop

Wrap sync_config, send_heartbeat, and queue.flush operations in separate
try/except to ensure daemon survives periodic operation exceptions. Critical
for unattended 24/7 pilot devices where a locked database or disk error would
otherwise kill the entire loop. Per-frame step() handling unchanged.
---
 edge/daemon.py       | 27 +++++++++++++++------------
 tests/test_daemon.py | 13 +++++++++++++
 2 files changed, 28 insertions(+), 12 deletions(-)

diff --git a/edge/daemon.py b/edge/daemon.py
index 408892b..7bd04a3 100644
--- a/edge/daemon.py
+++ b/edge/daemon.py
@@ -13,18 +13,21 @@ def run_loop(runtime, client, queue, *, clock, sleep, iterations=None,
             counters["frames"] += 1
         except Exception:
             counters["errors"] += 1
-        if sync_every and i % sync_every == 0:
-            cfg = client.fetch_config()
-            if cfg is not None:
-                from edge.zones import zones_from_config
-                runtime.apply_zones(zones_from_config(cfg))
-                counters["syncs"] += 1
-        if heartbeat_every and i % heartbeat_every == 0:
-            if client.send_heartbeat({"fps": 0.0, "frames": counters["frames"]}):
-                counters["beats"] += 1
-        if flush_every and i % flush_every == 0:
-            queue.flush()
-            counters["flushes"] += 1
+        try:
+            if sync_every and i % sync_every == 0:
+                cfg = client.fetch_config()
+                if cfg is not None:
+                    from edge.zones import zones_from_config
+                    runtime.apply_zones(zones_from_config(cfg))
+                    counters["syncs"] += 1
+            if heartbeat_every and i % heartbeat_every == 0:
+                if client.send_heartbeat({"fps": 0.0, "frames": counters["frames"]}):
+                    counters["beats"] += 1
+            if flush_every and i % flush_every == 0:
+                queue.flush()
+                counters["flushes"] += 1
+        except Exception:
+            counters["errors"] += 1
         if frame_interval:
             sleep(frame_interval)
     return counters
diff --git a/tests/test_daemon.py b/tests/test_daemon.py
index 1a675be..2d01880 100644
--- a/tests/test_daemon.py
+++ b/tests/test_daemon.py
@@ -41,3 +41,16 @@ def test_loop_survives_step_exception():
                         sync_every=99, heartbeat_every=99, flush_every=99)
     assert rt.steps == 3  # kept going past the exception
     assert counters["errors"] == 1
+
+
+def test_loop_survives_flush_exception():
+    class BoomQueue(_Queue):
+        def flush(self):
+            self.flushes += 1
+            raise RuntimeError("db locked")
+    rt = _Runtime()
+    counters = run_loop(rt, _Client(), BoomQueue(), clock=lambda: 0.0,
+                        sleep=lambda s: None, iterations=5,
+                        sync_every=99, heartbeat_every=99, flush_every=2)
+    assert rt.steps == 5          # loop kept running past the flush exceptions
+    assert counters["errors"] >= 1