#!/bin/bash
# Unraid MCP Server Development Script
# Safely manages server processes during development with accurate process detection
set -euo pipefail
# Configuration
DEFAULT_PORT=6970
PROJECT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
LOG_DIR="/tmp"
LOG_FILE="$LOG_DIR/unraid-mcp.log"
PID_FILE="$LOG_DIR/dev.pid"
# Ensure logs directory exists
mkdir -p "$LOG_DIR"
# All colors are now handled by Rich logging system
# Helper function for unified Rich logging
log() {
local message="$1"
local level="${2:-info}"
local indent="${3:-0}"
local file_timestamp="$(date +'%Y-%m-%d %H:%M:%S')"
# Use unified Rich logger for beautiful console output - escape single quotes
local escaped_message="${message//\'/\'\"\'\"\'}"
uv run python -c "from unraid_mcp.config.logging import log_with_level_and_indent; log_with_level_and_indent('$escaped_message', '$level', $indent)"
# File output without color
printf "[%s] %s\n" "$file_timestamp" "$message" >> "$LOG_FILE"
}
# Convenience functions for different log levels
log_error() { log "$1" "error" "${2:-0}"; }
log_warning() { log "$1" "warning" "${2:-0}"; }
log_success() { log "$1" "success" "${2:-0}"; }
log_info() { log "$1" "info" "${2:-0}"; }
log_status() { log "$1" "status" "${2:-0}"; }
# Rich header function
log_header() {
uv run python -c "from unraid_mcp.config.logging import log_header; log_header('$1')"
}
# Rich separator function
log_separator() {
uv run python -c "from unraid_mcp.config.logging import log_separator; log_separator()"
}
# Get port from environment or use default
get_port() {
local port="${UNRAID_MCP_PORT:-$DEFAULT_PORT}"
echo "$port"
}
# Write PID to file
write_pid_file() {
local pid=$1
echo "$pid" > "$PID_FILE"
}
# Read PID from file
read_pid_file() {
if [[ -f "$PID_FILE" ]]; then
cat "$PID_FILE" 2>/dev/null
fi
}
# Check if PID file contains valid running process
is_pid_valid() {
local pid=$1
[[ -n "$pid" ]] && [[ "$pid" =~ ^[0-9]+$ ]] && kill -0 "$pid" 2>/dev/null
}
# Clean up PID file
cleanup_pid_file() {
if [[ -f "$PID_FILE" ]]; then
rm -f "$PID_FILE"
log_info "🗑️ Cleaned up PID file"
fi
}
# Get PID from PID file if valid, otherwise return empty
get_valid_pid_from_file() {
local pid=$(read_pid_file)
if is_pid_valid "$pid"; then
echo "$pid"
else
# Clean up stale PID file
[[ -f "$PID_FILE" ]] && cleanup_pid_file
echo ""
fi
}
# Find processes using multiple detection methods
find_server_processes() {
local port=$(get_port)
local pids=()
# Method 0: Check PID file first (most reliable)
local pid_from_file=$(get_valid_pid_from_file)
if [[ -n "$pid_from_file" ]]; then
log_status "🔍 Found server PID from file: $pid_from_file"
pids+=("$pid_from_file")
fi
# Method 1: Command line pattern matching (fallback)
while IFS= read -r line; do
if [[ -n "$line" ]]; then
local pid=$(echo "$line" | awk '{print $2}')
# Add to pids if not already present
if [[ ! " ${pids[@]} " =~ " $pid " ]]; then
pids+=("$pid")
fi
fi
done < <(ps aux | grep -E 'python.*unraid.*mcp|python.*main\.py|uv run.*main\.py|uv run -m unraid_mcp' | grep -v grep | grep -v "$0")
# Method 2: Port binding verification (fallback)
if command -v lsof >/dev/null 2>&1; then
while IFS= read -r line; do
if [[ -n "$line" ]]; then
local pid=$(echo "$line" | awk '{print $2}')
# Add to pids if not already present
if [[ ! " ${pids[@]} " =~ " $pid " ]]; then
pids+=("$pid")
fi
fi
done < <(lsof -i ":$port" 2>/dev/null | grep LISTEN || true)
fi
# Method 3: Working directory verification for fallback methods
local verified_pids=()
for pid in "${pids[@]}"; do
# Skip if not a valid PID
if ! [[ "$pid" =~ ^[0-9]+$ ]]; then
continue
fi
# If this PID came from the PID file, it's already verified
if [[ "$pid" == "$pid_from_file" ]]; then
verified_pids+=("$pid")
continue
fi
# Verify other PIDs by working directory
if [[ -d "/proc/$pid" ]]; then
local pwd_info=""
if command -v pwdx >/dev/null 2>&1; then
pwd_info=$(pwdx "$pid" 2>/dev/null | cut -d' ' -f2- || echo "unknown")
else
pwd_info=$(readlink -f "/proc/$pid/cwd" 2>/dev/null || echo "unknown")
fi
# Verify it's running from our project directory or a parent directory
if [[ "$pwd_info" == "$PROJECT_DIR"* ]] || [[ "$pwd_info" == *"unraid-mcp"* ]]; then
verified_pids+=("$pid")
fi
fi
done
# Output final list
printf '%s\n' "${verified_pids[@]}" | grep -E '^[0-9]+$' || true
}
# Terminate a process gracefully, then forcefully if needed
terminate_process() {
local pid=$1
local name=${2:-"process"}
if ! kill -0 "$pid" 2>/dev/null; then
log_warning "⚠️ Process $pid ($name) already terminated"
return 0
fi
log_warning "🔄 Terminating $name (PID: $pid)..."
# Step 1: Graceful shutdown (SIGTERM)
log_info "→ Sending SIGTERM to PID $pid" 1
kill -TERM "$pid" 2>/dev/null || {
log_warning "⚠️ Failed to send SIGTERM (process may have died)" 2
return 0
}
# Step 2: Wait for graceful shutdown (5 seconds)
local count=0
while [[ $count -lt 5 ]]; do
if ! kill -0 "$pid" 2>/dev/null; then
log_success "✅ Process $pid terminated gracefully" 1
# Clean up PID file if this was our server process
local pid_from_file=$(read_pid_file)
if [[ "$pid" == "$pid_from_file" ]]; then
cleanup_pid_file
fi
return 0
fi
sleep 1
((count++))
log_info "⏳ Waiting for graceful shutdown... (${count}/5)" 2
done
# Step 3: Force kill (SIGKILL)
log_error "⚡ Graceful shutdown timeout, sending SIGKILL to PID $pid" 1
kill -KILL "$pid" 2>/dev/null || {
log_warning "⚠️ Failed to send SIGKILL (process may have died)" 2
return 0
}
# Step 4: Final verification
sleep 1
if kill -0 "$pid" 2>/dev/null; then
log_error "❌ Failed to terminate process $pid" 1
return 1
else
log_success "✅ Process $pid terminated forcefully" 1
# Clean up PID file if this was our server process
local pid_from_file=$(read_pid_file)
if [[ "$pid" == "$pid_from_file" ]]; then
cleanup_pid_file
fi
return 0
fi
}
# Stop all server processes
stop_servers() {
log_header "Server Shutdown"
log_error "🛑 Stopping existing server processes..."
local pids=($(find_server_processes))
if [[ ${#pids[@]} -eq 0 ]]; then
log_success "✅ No processes to stop"
return 0
fi
local failed=0
for pid in "${pids[@]}"; do
if ! terminate_process "$pid" "Unraid MCP Server"; then
((failed++))
fi
done
# Wait for ports to be released
local port=$(get_port)
log_info "⏳ Waiting for port $port to be released..."
local port_wait=0
while [[ $port_wait -lt 3 ]]; do
if ! lsof -i ":$port" >/dev/null 2>&1; then
log_success "✅ Port $port released" 1
break
fi
sleep 1
((port_wait++))
done
if [[ $failed -gt 0 ]]; then
log_error "⚠️ Failed to stop $failed process(es)"
return 1
else
log_success "✅ All processes stopped successfully"
return 0
fi
}
# Start the new modular server
start_modular_server() {
log_header "Modular Server Startup"
log_success "🚀 Starting modular server..."
cd "$PROJECT_DIR"
# Check if main.py exists in unraid_mcp/
if [[ ! -f "unraid_mcp/main.py" ]]; then
log_error "❌ unraid_mcp/main.py not found. Make sure modular server is implemented."
return 1
fi
# Clear the log file and add a startup marker to capture fresh logs
echo "=== Server Starting at $(date) ===" > "$LOG_FILE"
# Start server in background using module syntax
log_info "→ Executing: uv run -m unraid_mcp.main" 1
# Start server in new process group to isolate it from parent signals
setsid nohup uv run -m unraid_mcp.main >> "$LOG_FILE" 2>&1 &
local pid=$!
# Write PID to file
write_pid_file "$pid"
log_info "📝 Written PID $pid to file: $PID_FILE" 1
# Give it a moment to start and write some logs
sleep 3
# Check if it's still running
if kill -0 "$pid" 2>/dev/null; then
local port=$(get_port)
log_success "✅ Modular server started successfully (PID: $pid, Port: $port)"
log_info "📋 Process info: $(ps -p "$pid" -o pid,ppid,cmd --no-headers 2>/dev/null || echo 'Process info unavailable')" 1
# Auto-tail logs after successful start
echo ""
log_success "📄 Following server logs in real-time..."
log_info "ℹ️ Press Ctrl+C to stop following logs (server will continue running)" 1
log_separator
echo ""
# Set up signal handler for graceful exit from log following
trap 'handle_log_interrupt' SIGINT
# Start tailing from beginning of the fresh log file
tail -f "$LOG_FILE"
return 0
else
log_error "❌ Modular server failed to start"
cleanup_pid_file
log_warning "📄 Check $LOG_FILE for error details"
return 1
fi
}
# Start the original server
start_original_server() {
log_header "Original Server Startup"
log_success "🚀 Starting original server..."
cd "$PROJECT_DIR"
# Check if original server exists
if [[ ! -f "unraid_mcp_server.py" ]]; then
log_error "❌ unraid_mcp_server.py not found"
return 1
fi
# Clear the log file and add a startup marker to capture fresh logs
echo "=== Server Starting at $(date) ===" > "$LOG_FILE"
# Start server in background
log_info "→ Executing: uv run unraid_mcp_server.py" 1
# Start server in new process group to isolate it from parent signals
setsid nohup uv run unraid_mcp_server.py >> "$LOG_FILE" 2>&1 &
local pid=$!
# Write PID to file
write_pid_file "$pid"
log_info "📝 Written PID $pid to file: $PID_FILE" 1
# Give it a moment to start and write some logs
sleep 3
# Check if it's still running
if kill -0 "$pid" 2>/dev/null; then
local port=$(get_port)
log_success "✅ Original server started successfully (PID: $pid, Port: $port)"
log_info "📋 Process info: $(ps -p "$pid" -o pid,ppid,cmd --no-headers 2>/dev/null || echo 'Process info unavailable')" 1
# Auto-tail logs after successful start
echo ""
log_success "📄 Following server logs in real-time..."
log_info "ℹ️ Press Ctrl+C to stop following logs (server will continue running)" 1
log_separator
echo ""
# Set up signal handler for graceful exit from log following
trap 'handle_log_interrupt' SIGINT
# Start tailing from beginning of the fresh log file
tail -f "$LOG_FILE"
return 0
else
log_error "❌ Original server failed to start"
cleanup_pid_file
log_warning "📄 Check $LOG_FILE for error details"
return 1
fi
}
# Show usage information
show_usage() {
echo "Usage: $0 [OPTIONS]"
echo ""
echo "Development script for Unraid MCP Server"
echo ""
echo "OPTIONS:"
echo " (no args) Stop existing servers, start modular server, and tail logs"
echo " --old Stop existing servers, start original server, and tail logs"
echo " --kill Stop existing servers only (don't start new one)"
echo " --status Show status of running servers"
echo " --logs [N] Show last N lines of server logs (default: 50)"
echo " --tail Follow server logs in real-time (without restarting server)"
echo " --help, -h Show this help message"
echo ""
echo "ENVIRONMENT VARIABLES:"
echo " UNRAID_MCP_PORT Port for server (default: $DEFAULT_PORT)"
echo ""
echo "EXAMPLES:"
echo " ./dev.sh # Restart with modular server"
echo " ./dev.sh --old # Restart with original server"
echo " ./dev.sh --kill # Stop all servers"
echo " ./dev.sh --status # Check server status"
echo " ./dev.sh --logs # Show last 50 lines of logs"
echo " ./dev.sh --logs 100 # Show last 100 lines of logs"
echo " ./dev.sh --tail # Follow logs in real-time"
}
# Show server status
show_status() {
local port=$(get_port)
log_header "Server Status"
log_status "🔍 Server Status Check"
log_info "📁 Project Directory: $PROJECT_DIR" 1
log_info "📝 PID File: $PID_FILE" 1
log_info "🔌 Expected Port: $port" 1
echo ""
# Check PID file status
local pid_from_file=$(read_pid_file)
if [[ -n "$pid_from_file" ]]; then
if is_pid_valid "$pid_from_file"; then
log_success "✅ PID File: Contains valid PID $pid_from_file" 1
else
log_warning "⚠️ PID File: Contains stale PID $pid_from_file (process not running)" 1
fi
else
log_warning "🚫 PID File: Not found or empty" 1
fi
echo ""
local pids=($(find_server_processes))
if [[ ${#pids[@]} -eq 0 ]]; then
log_warning "🟡 Status: No servers running" 1
else
log_success "✅ Status: ${#pids[@]} server(s) running" 1
for pid in "${pids[@]}"; do
local cmd=$(ps -p "$pid" -o cmd --no-headers 2>/dev/null || echo "Command unavailable")
local source="process scan"
if [[ "$pid" == "$pid_from_file" ]]; then
source="PID file"
fi
log_success "PID $pid ($source): $cmd" 2
done
fi
# Check port binding
if command -v lsof >/dev/null 2>&1; then
local port_info=$(lsof -i ":$port" 2>/dev/null | grep LISTEN || echo "")
if [[ -n "$port_info" ]]; then
log_success "Port $port: BOUND" 1
echo "$port_info" | while IFS= read -r line; do
log_info "$line" 2
done
else
log_warning "Port $port: FREE" 1
fi
fi
}
# Tail the server logs
tail_logs() {
local lines="${1:-50}"
log_info "📄 Tailing last $lines lines from server logs..."
if [[ ! -f "$LOG_FILE" ]]; then
log_error "❌ Log file not found: $LOG_FILE"
return 1
fi
echo ""
echo "=== Server Logs (last $lines lines) ==="
tail -n "$lines" "$LOG_FILE"
echo "=== End of Logs ===="
echo ""
}
# Handle SIGINT during log following
handle_log_interrupt() {
echo ""
log_info "📄 Stopped following logs. Server continues running in background."
log_info "💡 Use './dev.sh --status' to check server status" 1
log_info "💡 Use './dev.sh --tail' to resume following logs" 1
exit 0
}
# Follow server logs in real-time
follow_logs() {
log_success "📄 Following server logs in real-time..."
log_info "ℹ️ Press Ctrl+C to stop following logs"
if [[ ! -f "$LOG_FILE" ]]; then
log_error "❌ Log file not found: $LOG_FILE"
return 1
fi
# Set up signal handler for graceful exit
trap 'handle_log_interrupt' SIGINT
log_separator
echo ""
tail -f "$LOG_FILE"
}
# Main script logic
main() {
# Initialize log file
echo "=== Dev Script Started at $(date) ===" >> "$LOG_FILE"
case "${1:-}" in
--help|-h)
show_usage
;;
--status)
show_status
;;
--kill)
stop_servers
;;
--logs)
tail_logs "${2:-50}"
;;
--tail)
follow_logs
;;
--old)
if stop_servers; then
start_original_server
else
log_error "❌ Failed to stop existing servers"
exit 1
fi
;;
"")
if stop_servers; then
start_modular_server
else
log_error "❌ Failed to stop existing servers"
exit 1
fi
;;
*)
log_error "❌ Unknown option: $1"
show_usage
exit 1
;;
esac
}
# Run main function with all arguments
main "$@"