Merge pull request 'Feature Add: Resource usage graphs' (#6) from experimental into main

Reviewed-on: #6

pterodisbot.py

@@ -30,7 +30,13 @@ import configparser
 from datetime import datetime
 from typing import Dict, List, Optional, Tuple
 from pathlib import Path
-import generate_config
+import matplotlib
+matplotlib.use('Agg')  # Use non-interactive backend for server environments
+import matplotlib.pyplot as plt
+import matplotlib.dates as mdates
+from collections import deque
+import io
+from server_metrics_graphs import ServerMetricsGraphs, ServerMetricsManager
 
 # ==============================================
 # LOGGING SETUP
@@ -549,10 +555,11 @@ class PterodactylBot(commands.Bot):
         self.server_cache: Dict[str, dict] = {}  # Cache of server data from Pterodactyl
         self.embed_locations: Dict[str, Dict[str, int]] = {}  # Tracks where embeds are posted
         self.update_lock = asyncio.Lock()  # Prevents concurrent updates
-        self.embed_storage_path = Path(EMBED_LOCATIONS_FILE)  # File to store embed locations
+        self.embed_storage_path = Path(EMBED_LOCATIONS_FILE)  # File to store embed 
+        self.metrics_manager = ServerMetricsManager() # Data manager for metrics graphing system
         # Track previous server states and CPU usage to detect changes
         # Format: {server_id: (state, cpu_usage, last_force_update)}
-        self.previous_states: Dict[str, Tuple[str, float, Optional[float]]] = {}  
+        self.previous_states: Dict[str, Tuple[str, float, Optional[float]]] = {}
         logger.info("Initialized PterodactylBot instance with state tracking")
         
     async def setup_hook(self):
@@ -753,12 +760,12 @@ class PterodactylBot(commands.Bot):
         embed.add_field(name="🆔 Server ID", value=f"`{identifier}`", inline=True)
         
         if is_suspended:
-            embed.add_field(name="📊 Status", value="⛔ Suspended", inline=True)
+            embed.add_field(name="ℹ️ Status", value="⛔ `Suspended`", inline=True)
         else:
-            embed.add_field(name="📊 Status", value="✅ Active", inline=True)
+            embed.add_field(name="ℹ️ Status", value="✅ `Active`", inline=True)
             
         # Add resource usage if server is running
-        if current_state.lower() == "running":
+        if current_state.lower() != "offline":
             # Current usage
             cpu_usage = round(resource_attributes.get('resources', {}).get('cpu_absolute', 0), 2)
             memory_usage = round(resource_attributes.get('resources', {}).get('memory_bytes', 0) / (1024 ** 2), 2)
@@ -786,19 +793,19 @@ class PterodactylBot(commands.Bot):
             if uptime_ms > 0:
                 uptime_seconds = uptime_ms // 1000  # Convert ms to seconds
                 if uptime_seconds < 60:
-                    uptime_text = f"{uptime_seconds}s"
+                    uptime_text = f"`{uptime_seconds}s`"
                 elif uptime_seconds < 3600:
-                    uptime_text = f"{uptime_seconds // 60}m {uptime_seconds % 60}s"
+                    uptime_text = f"`{uptime_seconds // 60}m {uptime_seconds % 60}s`"
                 elif uptime_seconds < 86400:
                     hours = uptime_seconds // 3600
                     minutes = (uptime_seconds % 3600) // 60
-                    uptime_text = f"{hours}h {minutes}m"
+                    uptime_text = f"`{hours}h {minutes}m`"
                 else:
                     days = uptime_seconds // 86400
                     hours = (uptime_seconds % 86400) // 3600
-                    uptime_text = f"{days}d {hours}h"
+                    uptime_text = f"`{days}d {hours}h`"
             else:
-                uptime_text = "Just started"
+                uptime_text = "`Just started`"
             
             embed.add_field(name="⏱️ Uptime", value=uptime_text, inline=True)
             
@@ -806,22 +813,48 @@ class PterodactylBot(commands.Bot):
             usage_text = (
                 f"```properties\n"
                 f"CPU:    {cpu_usage:>8} / {format_limit(cpu_limit, ' %')}\n"
-                f"Memory: {memory_usage:>8} / {format_limit(memory_limit, ' MB')}\n"
+                f"Memory: {memory_usage:>8} / {format_limit(memory_limit, ' MiB')}\n"
-                f"Disk:   {disk_usage:>8} / {format_limit(disk_limit, ' MB')}\n"
+                f"Disk:   {disk_usage:>8} / {format_limit(disk_limit, ' MiB')}\n"
                 f"```"
             )
             
             embed.add_field(
-                name="📈 Resource Usage",
+                name="📊 Resource Usage",
                 value=usage_text,
                 inline=False
             )
             
             embed.add_field(
-                name="🌐 Network", 
+                name="Network In", 
-                value=f"⬇️ {network_rx} MB / ⬆️ {network_tx} MB", 
+                value=f"📥 `{network_rx} MiB`", 
-                inline=False
+                inline=True
             )
+
+            embed.add_field(
+                name="Network Out", 
+                value=f"📤 `{network_tx} MiB`", 
+                inline=True
+            )
+            
+            # Add graph images if available
+            server_graphs = self.metrics_manager.get_server_graphs(identifier)
+            if server_graphs and server_graphs.has_sufficient_data:
+                summary = server_graphs.get_data_summary()
+
+                graph_description = (
+                    f">>> `Data points: {summary['point_count']}/6`\n"
+                    f"`CPU trend: {summary['cpu_trend']} • Memory trend: {summary['memory_trend']}`"
+                )
+                
+                # Add a field explaining the graphs
+                embed.add_field(
+                    name="📈 Usage Trends (Last Minute)",
+                    value=graph_description,
+                    inline=False
+                )
+                
+                # Set graph images (these will be attached as files in the update_status method)
+                embed.set_image(url=f"attachment://metrics_graph_{identifier}.png")
         
         embed.set_footer(text="Last updated")
         
@@ -860,6 +893,10 @@ class PterodactylBot(commands.Bot):
                 # Update our local cache with fresh server data
                 self.server_cache = {server['attributes']['identifier']: server for server in servers}
                 logger.debug(f"Updated server cache with {len(servers)} servers")
+
+                # Clean up metrics for servers that no longer exist
+                active_server_ids = list(self.server_cache.keys())
+                self.metrics_manager.cleanup_old_servers(active_server_ids)
                 
                 # Variables to track our update statistics
                 update_count = 0  # Successful updates
@@ -885,7 +922,13 @@ class PterodactylBot(commands.Bot):
                         resources = await self.pterodactyl_api.get_server_resources(server_id)
                         current_state = resources.get('attributes', {}).get('current_state', 'offline')
                         cpu_usage = round(resources.get('attributes', {}).get('resources', {}).get('cpu_absolute', 0), 2)
-                        
+
+                        # Collect metrics data for running servers
+                        if current_state == 'running':
+                            memory_usage = round(resources.get('attributes', {}).get('resources', {}).get('memory_bytes', 0) / (1024 ** 2), 2)
+                            self.metrics_manager.add_server_data(server_id, server_name, cpu_usage, memory_usage)
+                            logger.debug(f"Added metrics data for {server_name}: CPU={cpu_usage}%, Memory={memory_usage}MB")
+                            
                         # Retrieve previous recorded state, CPU usage, and last force update time
                         prev_state, prev_cpu, last_force_update = self.previous_states.get(server_id, (None, 0, None))
                         
@@ -909,8 +952,8 @@ class PterodactylBot(commands.Bot):
                         
                         # 4. Force update every 10 minutes for running servers (for uptime counter)
                         elif (current_state == 'running' and 
-                              (last_force_update is None or 
+                            (last_force_update is None or 
-                               current_time - last_force_update >= 600)):  # 10 minutes = 600 seconds
+                            current_time - last_force_update >= 600)):  # 10 minutes = 600 seconds
                             logger.debug(f"Executing 10-minute force update for running server {server_name}")
                             needs_update = True
                             # Update the last force update time
@@ -929,7 +972,28 @@ class PterodactylBot(commands.Bot):
                                 
                             # Fetch and update the existing message
                             message = await channel.fetch_message(int(location['message_id']))
-                            await message.edit(embed=embed, view=view)
+                            
+                            # Check if server is transitioning to offline/stopping state
+                            # and remove image attachment if present
+                            files = []
+                            server_graphs = self.metrics_manager.get_server_graphs(server_id)
+                            
+                            # Only include graph images if server is running AND has sufficient data
+                            if (current_state == 'running' and 
+                                server_graphs and 
+                                server_graphs.has_sufficient_data):
+                                # Generate metrics graph
+                                combined_graph = server_graphs.generate_combined_graph()
+                                if combined_graph:
+                                    files.append(discord.File(combined_graph, filename=f"metrics_graph_{server_id}.png"))
+                                    logger.debug(f"Including metrics graph for running server {server_name}")
+                            else:
+                                # Server is offline/stopping - ensure no image is attached
+                                logger.debug(f"Server {server_name} is {current_state}, removing image attachment if present")
+                                # We'll update without files to remove any existing attachments
+                            
+                            # Update message with embed, view, and files (empty files list removes attachments)
+                            await message.edit(embed=embed, view=view, attachments=files)
                             update_count += 1
                             logger.debug(f"Updated status for {server_name}")
                             

requirements.txt

@@ -1,4 +1,5 @@
 discord.py>=2.3.0
 aiohttp>=3.8.0
 configparser>=5.3.0
-python-dotenv
+python-dotenv
+matplotlib

server_metrics_graphs.py

@@ -0,0 +1,472 @@
+"""
+Server Metrics Graphs Module for Pterodactyl Discord Bot
+
+This module provides graphing capabilities for server CPU and memory usage.
+Generates line graphs as PNG images for embedding in Discord messages.
+"""
+
+import matplotlib
+matplotlib.use('Agg')  # Use non-interactive backend for server environments
+import matplotlib.pyplot as plt
+import matplotlib.dates as mdates
+from collections import deque
+from datetime import datetime, timedelta
+from typing import Dict, Tuple, Optional
+import io
+import logging
+import math
+
+# Get the logger from the main bot module
+logger = logging.getLogger('pterodisbot')
+
+class ServerMetricsGraphs:
+    """
+    Manages CPU and memory usage graphs for individual servers.
+    
+    Features:
+    - Stores last 6 data points (1 minute of history at 10-second intervals)
+    - Generates PNG images of line graphs for Discord embedding
+    - Automatic data rotation (FIFO queue with max 6 points)
+    - Separate tracking for CPU percentage and memory MB usage
+    - Dynamic CPU scaling in 100% increments for multi-vCPU servers
+    - Clean graph styling optimized for Discord dark theme
+    """
+    
+    def __init__(self, server_id: str, server_name: str):
+        """
+        Initialize metrics tracking for a server.
+        
+        Args:
+            server_id: Pterodactyl server identifier
+            server_name: Human-readable server name
+        """
+        self.server_id = server_id
+        self.server_name = server_name
+        
+        # Use deque with maxlen=6 for automatic FIFO rotation
+        # Each entry is a tuple: (timestamp, cpu_percent, memory_mb)
+        self.data_points = deque(maxlen=6)
+        
+        # Track if we have enough data for meaningful graphs (at least 2 points)
+        self.has_sufficient_data = False
+        
+        logger.debug(f"Initialized metrics tracking for server {server_name} ({server_id})")
+    
+    def add_data_point(self, cpu_percent: float, memory_mb: float, timestamp: Optional[datetime] = None):
+        """
+        Add a new data point to the metrics history.
+        
+        Args:
+            cpu_percent: Current CPU usage percentage
+            memory_mb: Current memory usage in megabytes
+            timestamp: Optional timestamp, defaults to current time
+        """
+        if timestamp is None:
+            timestamp = datetime.now()
+        
+        # Add new data point (automatically rotates old data due to maxlen=6)
+        self.data_points.append((timestamp, cpu_percent, memory_mb))
+        
+        # Update sufficient data flag
+        self.has_sufficient_data = len(self.data_points) >= 2
+        
+        logger.debug(f"Added metrics data point for {self.server_name}: CPU={cpu_percent}%, Memory={memory_mb}MB")
+    
+    def _calculate_cpu_scale_limit(self, max_cpu_value: float) -> int:
+        """
+        Calculate appropriate CPU scale limit in 100% increments.
+        
+        Args:
+            max_cpu_value: Maximum CPU value in the dataset
+            
+        Returns:
+            Scale limit rounded up to nearest 100% increment
+        """
+        if max_cpu_value <= 100:
+            return 100
+        
+        # Round up to nearest 100% increment
+        # e.g., 150% -> 200%, 250% -> 300%, 350% -> 400%
+        return math.ceil(max_cpu_value / 100) * 100
+    
+    def generate_cpu_graph(self) -> Optional[io.BytesIO]:
+        """
+        Generate a CPU usage line graph as a PNG image.
+        
+        Returns:
+            BytesIO object containing PNG image data, or None if insufficient data
+        """
+        if not self.has_sufficient_data:
+            logger.debug(f"Insufficient data for CPU graph generation: {self.server_name}")
+            return None
+        
+        try:
+            # Extract timestamps and CPU data
+            timestamps = [point[0] for point in self.data_points]
+            cpu_values = [point[1] for point in self.data_points]
+            
+            # Calculate dynamic CPU scale limit
+            max_cpu = max(cpu_values)
+            cpu_scale_limit = self._calculate_cpu_scale_limit(max_cpu)
+            
+            # Create figure with dark theme styling
+            plt.style.use('dark_background')
+            fig, ax = plt.subplots(figsize=(8, 4), dpi=100)
+            fig.patch.set_facecolor('#2f3136')  # Discord dark theme background
+            ax.set_facecolor('#36393f')  # Slightly lighter for graph area
+            
+            # Plot CPU line with gradient fill
+            line = ax.plot(timestamps, cpu_values, color='#7289da', linewidth=2.5, marker='o', markersize=4)
+            ax.fill_between(timestamps, cpu_values, alpha=0.3, color='#7289da')
+            
+            # Customize axes with dynamic scaling
+            ax.set_ylabel('CPU Usage (%)', color='#ffffff', fontsize=10)
+            ax.set_ylim(0, cpu_scale_limit)
+            
+            # Add horizontal grid lines at 100% increments for better readability
+            for i in range(100, cpu_scale_limit + 1, 100):
+                ax.axhline(y=i, color='#ffffff', alpha=0.2, linestyle='--', linewidth=0.8)
+            
+            # Format time axis
+            ax.xaxis.set_major_formatter(mdates.DateFormatter('%H:%M:%S'))
+            ax.xaxis.set_major_locator(mdates.SecondLocator(interval=20))
+            plt.setp(ax.xaxis.get_majorticklabels(), rotation=45, ha='right', color='#ffffff', fontsize=8)
+            
+            # Style the graph
+            ax.tick_params(colors='#ffffff', labelsize=8)
+            ax.grid(True, alpha=0.3, color='#ffffff')
+            ax.spines['bottom'].set_color('#ffffff')
+            ax.spines['left'].set_color('#ffffff')
+            ax.spines['top'].set_visible(False)
+            ax.spines['right'].set_visible(False)
+            
+            # Add title with scale info for multi-vCPU servers
+            title = f'{self.server_name} - CPU Usage'
+            if cpu_scale_limit > 100:
+                estimated_vcpus = cpu_scale_limit // 100
+                title += f' (~{estimated_vcpus} vCPU cores)'
+            ax.set_title(title, color='#ffffff', fontsize=12, pad=20)
+            
+            # Tight layout to prevent label cutoff
+            plt.tight_layout()
+            
+            # Save to BytesIO
+            img_buffer = io.BytesIO()
+            plt.savefig(img_buffer, format='png', facecolor='#2f3136', edgecolor='none', 
+                       bbox_inches='tight', dpi=100)
+            img_buffer.seek(0)
+            
+            # Clean up matplotlib resources
+            plt.close(fig)
+            
+            logger.debug(f"Generated CPU graph for {self.server_name} (scale: 0-{cpu_scale_limit}%)")
+            return img_buffer
+            
+        except Exception as e:
+            logger.error(f"Failed to generate CPU graph for {self.server_name}: {str(e)}")
+            plt.close('all')  # Clean up any remaining figures
+            return None
+    
+    def generate_memory_graph(self) -> Optional[io.BytesIO]:
+        """
+        Generate a memory usage line graph as a PNG image.
+        
+        Returns:
+            BytesIO object containing PNG image data, or None if insufficient data
+        """
+        if not self.has_sufficient_data:
+            logger.debug(f"Insufficient data for memory graph generation: {self.server_name}")
+            return None
+        
+        try:
+            # Extract timestamps and memory data
+            timestamps = [point[0] for point in self.data_points]
+            memory_values = [point[2] for point in self.data_points]
+            
+            # Create figure with dark theme styling
+            plt.style.use('dark_background')
+            fig, ax = plt.subplots(figsize=(8, 4), dpi=100)
+            fig.patch.set_facecolor('#2f3136')  # Discord dark theme background
+            ax.set_facecolor('#36393f')  # Slightly lighter for graph area
+            
+            # Plot memory line with gradient fill
+            line = ax.plot(timestamps, memory_values, color='#43b581', linewidth=2.5, marker='o', markersize=4)
+            ax.fill_between(timestamps, memory_values, alpha=0.3, color='#43b581')
+            
+            # Customize axes
+            ax.set_ylabel('Memory Usage (MB)', color='#ffffff', fontsize=10)
+            ax.set_ylim(0, max(memory_values) * 1.1)  # Dynamic scaling with 10% padding
+            
+            # Format time axis
+            ax.xaxis.set_major_formatter(mdates.DateFormatter('%H:%M:%S'))
+            ax.xaxis.set_major_locator(mdates.SecondLocator(interval=20))
+            plt.setp(ax.xaxis.get_majorticklabels(), rotation=45, ha='right', color='#ffffff', fontsize=8)
+            
+            # Style the graph
+            ax.tick_params(colors='#ffffff', labelsize=8)
+            ax.grid(True, alpha=0.3, color='#ffffff')
+            ax.spines['bottom'].set_color('#ffffff')
+            ax.spines['left'].set_color('#ffffff')
+            ax.spines['top'].set_visible(False)
+            ax.spines['right'].set_visible(False)
+            
+            # Add title
+            ax.set_title(f'{self.server_name} - Memory Usage', color='#ffffff', fontsize=12, pad=20)
+            
+            # Tight layout to prevent label cutoff
+            plt.tight_layout()
+            
+            # Save to BytesIO
+            img_buffer = io.BytesIO()
+            plt.savefig(img_buffer, format='png', facecolor='#2f3136', edgecolor='none', 
+                       bbox_inches='tight', dpi=100)
+            img_buffer.seek(0)
+            
+            # Clean up matplotlib resources
+            plt.close(fig)
+            
+            logger.debug(f"Generated memory graph for {self.server_name}")
+            return img_buffer
+            
+        except Exception as e:
+            logger.error(f"Failed to generate memory graph for {self.server_name}: {str(e)}")
+            plt.close('all')  # Clean up any remaining figures
+            return None
+
+    def generate_combined_graph(self) -> Optional[io.BytesIO]:
+        """
+        Generate a combined CPU and memory usage graph as a PNG image.
+        
+        Returns:
+            BytesIO object containing PNG image data, or None if insufficient data
+        """
+        if not self.has_sufficient_data:
+            logger.debug(f"Insufficient data for combined graph generation: {self.server_name}")
+            return None
+        
+        try:
+            # Extract data
+            timestamps = [point[0] for point in self.data_points]
+            cpu_values = [point[1] for point in self.data_points]
+            memory_values = [point[2] for point in self.data_points]
+            
+            # Calculate dynamic CPU scale limit
+            max_cpu = max(cpu_values)
+            cpu_scale_limit = self._calculate_cpu_scale_limit(max_cpu)
+            
+            # Create figure with two subplots
+            plt.style.use('dark_background')
+            fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(8, 6), dpi=100, sharex=True)
+            fig.patch.set_facecolor('#2f3136')
+            
+            # CPU subplot
+            ax1.set_facecolor('#36393f')
+            ax1.plot(timestamps, cpu_values, color='#7289da', linewidth=2.5, marker='o', markersize=4)
+            ax1.fill_between(timestamps, cpu_values, alpha=0.3, color='#7289da')
+            ax1.set_ylabel('CPU Usage (%)', color='#ffffff', fontsize=10)
+            ax1.set_ylim(0, cpu_scale_limit)
+            ax1.tick_params(colors='#ffffff', labelsize=8)
+            ax1.grid(True, alpha=0.3, color='#ffffff')
+            
+            # Add horizontal grid lines at 100% increments for CPU subplot
+            for i in range(100, cpu_scale_limit + 1, 100):
+                ax1.axhline(y=i, color='#ffffff', alpha=0.2, linestyle='--', linewidth=0.8)
+            
+            # Title with vCPU info if applicable
+            title = f'{self.server_name} - Resource Usage'
+            if cpu_scale_limit > 100:
+                estimated_vcpus = cpu_scale_limit // 100
+                title += f' (~{estimated_vcpus} vCPU cores)'
+            ax1.set_title(title, color='#ffffff', fontsize=12)
+            
+            # Memory subplot
+            ax2.set_facecolor('#36393f')
+            ax2.plot(timestamps, memory_values, color='#43b581', linewidth=2.5, marker='o', markersize=4)
+            ax2.fill_between(timestamps, memory_values, alpha=0.3, color='#43b581')
+            ax2.set_ylabel('Memory (MB)', color='#ffffff', fontsize=10)
+            ax2.set_ylim(0, max(memory_values) * 1.1)
+            ax2.tick_params(colors='#ffffff', labelsize=8)
+            ax2.grid(True, alpha=0.3, color='#ffffff')
+            
+            # Format time axis (only on bottom subplot)
+            ax2.xaxis.set_major_formatter(mdates.DateFormatter('%H:%M:%S'))
+            ax2.xaxis.set_major_locator(mdates.SecondLocator(interval=20))
+            plt.setp(ax2.xaxis.get_majorticklabels(), rotation=45, ha='right', color='#ffffff', fontsize=8)
+            
+            # Style both subplots
+            for ax in [ax1, ax2]:
+                ax.spines['bottom'].set_color('#ffffff')
+                ax.spines['left'].set_color('#ffffff')
+                ax.spines['top'].set_visible(False)
+                ax.spines['right'].set_visible(False)
+            
+            plt.tight_layout()
+            
+            # Save to BytesIO
+            img_buffer = io.BytesIO()
+            plt.savefig(img_buffer, format='png', facecolor='#2f3136', edgecolor='none', 
+                    bbox_inches='tight', dpi=100)
+            img_buffer.seek(0)
+            
+            plt.close(fig)
+            
+            logger.debug(f"Generated combined graph for {self.server_name} (CPU scale: 0-{cpu_scale_limit}%)")
+            return img_buffer
+            
+        except Exception as e:
+            logger.error(f"Failed to generate combined graph for {self.server_name}: {str(e)}")
+            plt.close('all')
+            return None
+    
+    def get_data_summary(self) -> Dict[str, any]:
+        """
+        Get summary statistics for the current data points.
+        
+        Returns:
+            Dictionary containing data point count, latest values, and trends
+        """
+        if not self.data_points:
+            return {
+                'point_count': 0,
+                'has_data': False,
+                'latest_cpu': 0,
+                'latest_memory': 0
+            }
+        
+        # Get latest values
+        latest_point = self.data_points[-1]
+        latest_cpu = latest_point[1]
+        latest_memory = latest_point[2]
+        
+        # Calculate CPU scale info
+        max_cpu = max(point[1] for point in self.data_points)
+        cpu_scale_limit = self._calculate_cpu_scale_limit(max_cpu)
+        estimated_vcpus = cpu_scale_limit // 100
+        
+        # Calculate trends if we have multiple points
+        cpu_trend = 'stable'
+        memory_trend = 'stable'
+        
+        if len(self.data_points) >= 2:
+            first_point = self.data_points[0]
+            cpu_change = latest_cpu - first_point[1]
+            memory_change = latest_memory - first_point[2]
+            
+            # Determine trends (>5% change considered significant)
+            if abs(cpu_change) > 5:
+                cpu_trend = 'increasing' if cpu_change > 0 else 'decreasing'
+            
+            if abs(memory_change) > 50:  # 50MB change threshold
+                memory_trend = 'increasing' if memory_change > 0 else 'decreasing'
+        
+        return {
+            'point_count': len(self.data_points),
+            'has_data': self.has_sufficient_data,
+            'latest_cpu': latest_cpu,
+            'latest_memory': latest_memory,
+            'cpu_trend': cpu_trend,
+            'memory_trend': memory_trend,
+            'cpu_scale_limit': cpu_scale_limit,
+            'estimated_vcpus': estimated_vcpus,
+            'time_span_minutes': len(self.data_points) * 10 / 60  # Convert to minutes
+        }
+
+
+class ServerMetricsManager:
+    """
+    Global manager for all server metrics graphs.
+    
+    Handles:
+    - Creation and cleanup of ServerMetricsGraphs instances
+    - Bulk operations across all tracked servers
+    - Memory management for graph storage
+    """
+    
+    def __init__(self):
+        """Initialize the metrics manager."""
+        self.server_graphs: Dict[str, ServerMetricsGraphs] = {}
+        logger.info("Initialized ServerMetricsManager")
+    
+    def get_or_create_server_graphs(self, server_id: str, server_name: str) -> ServerMetricsGraphs:
+        """
+        Get existing ServerMetricsGraphs instance or create a new one.
+        
+        Args:
+            server_id: Pterodactyl server identifier
+            server_name: Human-readable server name
+            
+        Returns:
+            ServerMetricsGraphs instance for the specified server
+        """
+        if server_id not in self.server_graphs:
+            self.server_graphs[server_id] = ServerMetricsGraphs(server_id, server_name)
+            logger.debug(f"Created new metrics graphs for server {server_name}")
+        
+        return self.server_graphs[server_id]
+    
+    def add_server_data(self, server_id: str, server_name: str, cpu_percent: float, memory_mb: float):
+        """
+        Add data point to a server's metrics tracking.
+        
+        Args:
+            server_id: Pterodactyl server identifier
+            server_name: Human-readable server name
+            cpu_percent: Current CPU usage percentage
+            memory_mb: Current memory usage in megabytes
+        """
+        graphs = self.get_or_create_server_graphs(server_id, server_name)
+        graphs.add_data_point(cpu_percent, memory_mb)
+    
+    def remove_server(self, server_id: str):
+        """
+        Remove a server from metrics tracking.
+        
+        Args:
+            server_id: Pterodactyl server identifier to remove
+        """
+        if server_id in self.server_graphs:
+            del self.server_graphs[server_id]
+            logger.debug(f"Removed metrics tracking for server {server_id}")
+    
+    def get_server_graphs(self, server_id: str) -> Optional[ServerMetricsGraphs]:
+        """
+        Get ServerMetricsGraphs instance for a specific server.
+        
+        Args:
+            server_id: Pterodactyl server identifier
+            
+        Returns:
+            ServerMetricsGraphs instance or None if not found
+        """
+        return self.server_graphs.get(server_id)
+    
+    def cleanup_old_servers(self, active_server_ids: list):
+        """
+        Remove tracking for servers that no longer exist.
+        
+        Args:
+            active_server_ids: List of currently active server IDs
+        """
+        servers_to_remove = []
+        for server_id in self.server_graphs:
+            if server_id not in active_server_ids:
+                servers_to_remove.append(server_id)
+        
+        for server_id in servers_to_remove:
+            self.remove_server(server_id)
+        
+        if servers_to_remove:
+            logger.info(f"Cleaned up metrics for {len(servers_to_remove)} inactive servers")
+    
+    def get_summary(self) -> Dict[str, any]:
+        """
+        Get summary of all tracked servers.
+        
+        Returns:
+            Dictionary with tracking statistics
+        """
+        return {
+            'total_servers': len(self.server_graphs),
+            'servers_with_data': sum(1 for graphs in self.server_graphs.values() if graphs.has_sufficient_data),
+            'total_data_points': sum(len(graphs.data_points) for graphs in self.server_graphs.values())
+        }