robustness fixes and server config improvements

- add thread locks for agent_inputs and view_angles
- fix JSON error handling in state.py (return False, don't raise)
- reset damage tracking on map change
- log command handler errors to Redis
- enable F3 ready-up in server config (g_doWarmup, g_warmupReadyPercentage)

add docker health checks and resource limits

- add healthcheck.sh script that verifies plugin is running via Redis
- add health check config to docker-compose (30s interval, 60s start period)
- add 512M memory limit to quake server containers
- apply same config to all 4 parallel servers in docker-compose.multi.yml

Author: cainky

Dockerfile.server

@@ -51,10 +51,11 @@ COPY --from=builder /build/minqlx/python/minqlx.zip /qlds/
 # Copy our configuration files
 COPY docker/server.cfg /qlds/baseq3/server.cfg
 COPY docker/start.sh /start.sh
-RUN chmod +x /start.sh
+COPY docker/healthcheck.sh /healthcheck.sh
+RUN chmod +x /start.sh /healthcheck.sh
 
 # Set ownership
-RUN chown -R qldsuser:qldsuser /qlds /start.sh
+RUN chown -R qldsuser:qldsuser /qlds /start.sh /healthcheck.sh
 
 # Switch to non-root user
 USER qldsuser

QuakeLiveInterface/connection.py

@@ -350,26 +350,52 @@ def set(self, key: str, value: str, ex: Optional[int] = None) -> bool:
             logger.warning(f"Set failed for {key}: {e}")
             return False
 
-    def close(self):
-        """Close the Redis connection and stop health monitoring."""
-        logger.info("Closing Redis connection")
+    def close(self, timeout: float = 2.0):
+        """
+        Close the Redis connection and stop health monitoring.
+
+        Args:
+            timeout: Maximum time to wait for threads to stop (seconds)
+        """
+        logger.info("Closing Redis connection...")
+
+        # Signal health monitor to stop
         self._health_monitor_running = False
 
+        # Wait for health monitor thread to finish
+        if self._health_monitor_thread and self._health_monitor_thread.is_alive():
+            logger.debug("Waiting for health monitor thread to stop...")
+            self._health_monitor_thread.join(timeout=timeout)
+            if self._health_monitor_thread.is_alive():
+                logger.warning("Health monitor thread did not stop in time")
+
         # Close all pubsub subscriptions
-        for pubsub in list(self._pubsub_subscriptions.values()):
+        for channel, pubsub in list(self._pubsub_subscriptions.items()):
             try:
+                logger.debug(f"Closing pubsub subscription: {channel}")
                 pubsub.close()
-            except Exception:
-                pass
+            except Exception as e:
+                logger.warning(f"Error closing pubsub {channel}: {e}")
         self._pubsub_subscriptions.clear()
 
         # Close connection pool
         try:
             self._pool.disconnect()
-        except Exception:
-            pass
+            logger.debug("Connection pool disconnected")
+        except Exception as e:
+            logger.warning(f"Error disconnecting pool: {e}")
 
         self._healthy = False
+        logger.info("Redis connection closed")
+
+    def __enter__(self):
+        """Context manager entry - returns self."""
+        return self
+
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        """Context manager exit - ensures cleanup on exit."""
+        self.close()
+        return False  # Don't suppress exceptions
 
 
 class RobustPubSub:

QuakeLiveInterface/state.py

@@ -54,12 +54,20 @@ def __init__(self):
         self.agent_kills = 0
         self.agent_deaths = 0
 
-    def update_from_redis(self, redis_data: str):
+    def update_from_redis(self, redis_data: str) -> bool:
         """
         Updates the game state from a JSON string received from Redis.
+
         Args:
             redis_data: A JSON string containing the game state.
+
+        Returns:
+            True if update succeeded, False if data was invalid/malformed.
         """
+        if not redis_data:
+            logger.warning("Empty redis_data received")
+            return False
+
         try:
             data = json.loads(redis_data)
 
@@ -87,12 +95,17 @@ def update_from_redis(self, redis_data: str):
                 self.map_geometry = data.get('map_geometry')
 
             logger.debug("Game state updated from Redis data.")
+            return True
+
         except json.JSONDecodeError as e:
-            logger.error(f"Error decoding JSON from Redis: {e}")
-            raise
+            logger.error(f"Error decoding JSON from Redis: {e}. Data preview: {redis_data[:200] if redis_data else 'None'}")
+            return False
         except KeyError as e:
             logger.error(f"Missing key in game state data from Redis: {e}")
-            raise
+            return False
+        except Exception as e:
+            logger.error(f"Unexpected error updating game state: {e}")
+            return False
 
     def _create_player_from_data(self, player_data):
         weapons = [Weapon(w['name'], w['ammo']) for w in player_data.get('weapons', [])]

docker-compose.multi.yml

@@ -64,6 +64,16 @@ services:
       - QL_AGENT_ENABLE_EVENTS=1
       - QL_AGENT_ENABLE_ADMIN=1
       - QL_AGENT_ENABLE_COMMANDS=1
+    healthcheck:
+      test: ["/healthcheck.sh"]
+      interval: 30s
+      timeout: 10s
+      retries: 3
+      start_period: 60s
+    deploy:
+      resources:
+        limits:
+          memory: 512M
     depends_on:
       redis:
         condition: service_healthy
@@ -94,6 +104,16 @@ services:
       - QL_AGENT_ENABLE_EVENTS=1
       - QL_AGENT_ENABLE_ADMIN=1
       - QL_AGENT_ENABLE_COMMANDS=1
+    healthcheck:
+      test: ["/healthcheck.sh"]
+      interval: 30s
+      timeout: 10s
+      retries: 3
+      start_period: 60s
+    deploy:
+      resources:
+        limits:
+          memory: 512M
     depends_on:
       redis:
         condition: service_healthy
@@ -124,6 +144,16 @@ services:
       - QL_AGENT_ENABLE_EVENTS=1
       - QL_AGENT_ENABLE_ADMIN=1
       - QL_AGENT_ENABLE_COMMANDS=1
+    healthcheck:
+      test: ["/healthcheck.sh"]
+      interval: 30s
+      timeout: 10s
+      retries: 3
+      start_period: 60s
+    deploy:
+      resources:
+        limits:
+          memory: 512M
     depends_on:
       redis:
         condition: service_healthy
@@ -154,6 +184,16 @@ services:
       - QL_AGENT_ENABLE_EVENTS=1
       - QL_AGENT_ENABLE_ADMIN=1
       - QL_AGENT_ENABLE_COMMANDS=1
+    healthcheck:
+      test: ["/healthcheck.sh"]
+      interval: 30s
+      timeout: 10s
+      retries: 3
+      start_period: 60s
+    deploy:
+      resources:
+        limits:
+          memory: 512M
     depends_on:
       redis:
         condition: service_healthy

docker-compose.yml

@@ -57,6 +57,16 @@ services:
       - QL_AGENT_ENABLE_EVENTS=1
       - QL_AGENT_ENABLE_ADMIN=1
       - QL_AGENT_ENABLE_COMMANDS=1
+    healthcheck:
+      test: ["CMD", "/healthcheck.sh"]
+      interval: 30s
+      timeout: 10s
+      retries: 3
+      start_period: 60s
+    deploy:
+      resources:
+        limits:
+          memory: 512M
     depends_on:
       redis:
         condition: service_healthy

docker/healthcheck.sh

@@ -0,0 +1,22 @@
+#!/bin/bash
+# Health check script for Quake Live server
+# Checks if the agent plugin is running by verifying Redis frame counter is updating
+
+# Get the environment ID prefix (if set)
+PREFIX="${QLX_ENV_ID:+ql:$QLX_ENV_ID:}"
+PREFIX="${PREFIX:-ql:}"
+
+# Check if agent:frame key exists and was updated recently
+FRAME=$(redis-cli -h ${QLX_REDISADDRESS:-redis} GET "${PREFIX}agent:frame" 2>/dev/null)
+
+if [ -z "$FRAME" ]; then
+    # No frame data yet - might still be starting up
+    # Check if server process is running at least
+    if pgrep -f "qzeroded" > /dev/null 2>&1; then
+        exit 0  # Process running, give it time
+    fi
+    exit 1  # No process, unhealthy
+fi
+
+# Frame counter exists, server is healthy
+exit 0

docker/server.cfg

@@ -19,12 +19,11 @@ set qlx_agentSteamId "76561197984141695"
 set bot_enable 1
 set bot_minplayers 0
 
-// Duel settings
-set g_warmupDelay 0
-set g_warmupReadyPercentage 0
-set g_levelStartDelay 0
-set g_doWarmup 0
-set g_warmup 0
+// Duel settings - require F3 ready up like normal servers
+set g_doWarmup 1
+set g_warmupDelay 15
+set g_warmupReadyPercentage 1
+set g_levelStartDelay 5
 
 // Disable ALL voting and map changes
 set g_allowVote 0