vibego 0.2.58__py3-none-any.whl → 1.0.10__py3-none-any.whl

scripts/start.sh

@@ -3,6 +3,9 @@ set -eo pipefail
 
 ROOT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
 MASTER_CONFIG_ROOT="${MASTER_CONFIG_ROOT:-$HOME/.config/vibego}"
+RUNTIME_DIR="${VIBEGO_RUNTIME_ROOT:-$MASTER_CONFIG_ROOT/runtime}"
+VENV_DIR="$RUNTIME_DIR/.venv"
+LEGACY_VENV_DIR="$ROOT_DIR/.venv"
 STATE_DIR="$MASTER_CONFIG_ROOT/state"
 LOG_DIR="$MASTER_CONFIG_ROOT/logs"
 LOCK_FILE="$STATE_DIR/master_restart.lock"
@@ -10,8 +13,8 @@ START_LOG="$LOG_DIR/start.log"
 CODEX_STAMP_FILE="$STATE_DIR/npm_codex_install.stamp"
 CODEX_INSTALL_TTL="${CODEX_INSTALL_TTL:-86400}"
 
-# 统一重启信号文件路径：使用配置目录而非代码目录
-# 这样 pipx 安装的 master 和源码运行的 master 可以共享同一个信号文件
+# Unified restart signal file path: use the configuration directory instead of the code directory
+# In this way, the master installed by pipx and the master run from the source code can share the same signal file.
 export MASTER_RESTART_SIGNAL_PATH="$STATE_DIR/restart_signal.json"
 export LOG_ROOT="${LOG_ROOT:-$LOG_DIR}"
 if [[ -z "${LOG_FILE:-}" ]]; then
@@ -42,34 +45,35 @@ cd "$ROOT_DIR"
 
 mkdir -p "$(dirname "$LOCK_FILE")"
 mkdir -p "$(dirname "$START_LOG")"
+mkdir -p "$RUNTIME_DIR"
 touch "$START_LOG"
 exec >>"$START_LOG"
 exec 2>&1
 
-log_info "start.sh 启动，pid=$$"
+log_info "start.sh start, pid=$$"
 
 if [[ -f "$LOCK_FILE" ]]; then
-  log_error "已有 start.sh 在执行，跳过本次启动。"
+  log_error "Already have start.sh During execution, skip this startup."
   exit 1
 fi
 
 printf '%d\n' $$ > "$LOCK_FILE"
 
-log_info "锁文件已创建：$LOCK_FILE"
+log_info "Lock file created:$LOCK_FILE"
 
 ensure_codex_installed() {
   local need_install=1
   local now
   local codex_bin
   if ! command -v npm >/dev/null 2>&1; then
-    log_error "未检测到 npm，可执行文件缺失，跳过 @openai/codex 全局安装"
+    log_error "npm not detected, executable missing, skipped @openai/codex Global installation"
     return
   fi
 
-  log_info "检测到 npm 版本：$(npm --version)"
+  log_info "npm version detected:$(npm --version)"
 
   if [[ ! "$CODEX_INSTALL_TTL" =~ ^[0-9]+$ ]]; then
-    log_error "CODEX_INSTALL_TTL 非法值：$CODEX_INSTALL_TTL，回退为 86400 秒"
+    log_error "CODEX_INSTALL_TTL Illegal value:$CODEX_INSTALL_TTL, Fallback to 86400 seconds"
     CODEX_INSTALL_TTL=86400
   fi
 
@@ -99,11 +103,11 @@ ensure_codex_installed() {
   fi
 
   if (( need_install )); then
-    log_info "开始执行 npm install -g @openai/codex@latest"
+    log_info "Start executing npm install -g @openai/codex@latest"
     if npm install -g @openai/codex@latest; then
       now=$(date +%s)
       printf '%s\n' "$now" > "$CODEX_STAMP_FILE"
-      log_info "npm install -g @openai/codex@latest 成功"
+      log_info "npm install -g @openai/codex@latest success"
     else
       local status=$?
       log_error "npm install -g @openai/codex@latest failed (exit code ${status}); continuing startup"
@@ -114,8 +118,9 @@ ensure_codex_installed() {
 ensure_codex_installed
 
 select_python_binary() {
-  # 选择满足 CPython <=3.12 的解释器，默认禁用 3.13（pydantic-core 在 pipx 基础环境下无兼容轮子）
+  # Select a compatible CPython version; defaults accept 3.9-3.14 and can be overridden via env variables.
   local allow_py313="${VIBEGO_ALLOW_PY313:-}"
+  local supported_max_minor="${VIBEGO_MAX_MINOR:-14}"
   local candidates=()
   local chosen=""
   local name
@@ -141,121 +146,127 @@ select_python_binary() {
     local major="${version_raw%%.*}"
     local minor="${version_raw#*.}"
     if [[ "$major" != "3" ]]; then
-      log_line "跳过 ${name} (版本 ${version_raw})：非 CPython 3.x" >&2
+      log_line "jump over ${name} (Version ${version_raw}): Non-CPython 3.x" >&2
       continue
     fi
     local explicit_override=0
     if [[ -n "${VIBEGO_PYTHON:-}" && "$name" == "$VIBEGO_PYTHON" ]]; then
       explicit_override=1
     fi
-    if [[ "$minor" =~ ^[0-9]+$ ]] && (( minor == 13 )) && [[ "$allow_py313" != "1" ]] && (( explicit_override == 0 )); then
-      log_line "跳过 ${name} (版本 ${version_raw})：默认禁用 Python 3.13，可设置 VIBEGO_ALLOW_PY313=1 覆盖" >&2
-      continue
+    if [[ "$minor" =~ ^[0-9]+$ ]] && (( minor == 13 )) && (( explicit_override == 0 )); then
+      if [[ "$allow_py313" == "0" ]]; then
+        log_line "Skip ${name} (version ${version_raw}): disabled explicitly by VIBEGO_ALLOW_PY313=0" >&2
+        continue
+      fi
+      log_line "Detected ${name} (version ${version_raw}): Python 3.13 accepted by default; use VIBEGO_ALLOW_PY313=1 to prefer or 0 to disable" >&2
     fi
-    if [[ "$minor" =~ ^[0-9]+$ ]] && (( minor > 13 )); then
-      log_line "跳过 ${name} (版本 ${version_raw})：高于 3.13" >&2
+    if [[ "$minor" =~ ^[0-9]+$ ]] && (( minor > supported_max_minor )) && (( explicit_override == 0 )); then
+      log_line "Skip ${name} (version ${version_raw}): above supported ceiling 3.${supported_max_minor}; override with VIBEGO_MAX_MINOR if needed" >&2
       continue
     fi
     if [[ "$minor" =~ ^[0-9]+$ ]] && (( minor < 9 )); then
-      log_line "跳过 ${name} (版本 ${version_raw})：低于 3.9，可能缺少官方轮子" >&2
+      log_line "jump over ${name} (Version ${version_raw}): less than 3.9, May be missing official wheels" >&2
       continue
     fi
+    if [[ "$minor" =~ ^[0-9]+$ ]] && (( minor >= 14 )); then
+      log_line "Detected ${name} (version ${version_raw}): ensure dependencies support this Python version" >&2
+    fi
     chosen="$name"
-    log_line "使用 Python 解释器：${chosen} (版本 ${version_raw})" >&2
+    log_line "Using the Python interpreter:${chosen} (Version ${version_raw})" >&2
     break
   done
 
   if [[ -z "$chosen" ]]; then
-    log_error "未找到满足 <=3.13 的 Python 解释器，可通过设置 VIBEGO_PYTHON 指定路径"
+    log_error "no satisfaction found <=3.13 The Python interpreter can be set by VIBEGO_PYTHON Specify path"
     exit 1
   fi
 
   printf '%s' "$chosen"
 }
 
-# 检查Python依赖是否已安装完整
+# Check whether Python dependencies are installed completely
 check_deps_installed() {
-  # 检查虚拟环境是否存在
-  if [[ ! -d "$ROOT_DIR/.venv" ]]; then
-    log_info "虚拟环境不存在，需要初始化"
+  # Check if the virtual environment exists
+  if [[ ! -d "$VENV_DIR" ]]; then
+    log_info "The virtual environment does not exist and needs to be initialized."
     return 1
   fi
 
-  # 检查虚拟环境的Python解释器
-  if [[ ! -x "$ROOT_DIR/.venv/bin/python" ]]; then
-    log_info "虚拟环境Python解释器缺失"
+  # Check the virtual environment's Python interpreter
+  if [[ ! -x "$VENV_DIR/bin/python" ]]; then
+    log_info "Virtual environment Python interpreter missing"
     return 1
   fi
 
-  # 激活虚拟环境并检查关键依赖包
-  # aiogram: Telegram Bot框架
-  # aiohttp: 异步HTTP客户端
-  # aiosqlite: 异步SQLite数据库
-  if ! "$ROOT_DIR/.venv/bin/python" -c "import aiogram, aiohttp, aiosqlite" 2>/dev/null; then
-    log_info "关键依赖包缺失或损坏"
+  # Activate the virtual environment and check key dependency packages
+  # aiogram: Telegram Botframe
+  # aiohttp: Asynchronous HTTP client
+  # aiosqlite: Asynchronous SQLite database
+  if ! "$VENV_DIR/bin/python" -c "import aiogram, aiohttp, aiosqlite" 2>/dev/null; then
+    log_info "Key dependency packages are missing or damaged"
     return 1
   fi
 
-  log_info "依赖检查通过，虚拟环境完整"
+  log_info "The dependency check passed and the virtual environment is complete"
   return 0
 }
 
-# 清理旧 master 进程的健壮函数（改进版：支持 PID 文件 + pgrep 双保险）
+# Robust function to clean up old master processes (improved version: supports PID files + pgrep Double insurance)
 cleanup_old_master() {
-  local max_wait=10  # 最多等待10秒优雅退出
+  local max_wait=10  # Wait up to 10 seconds to exit gracefully
   local waited=0
   local old_pids=""
   local master_pid_file="$STATE_DIR/master.pid"
 
-  # 方案1：优先从 PID 文件读取
+  # Option 1: Read from the PID file first
   if [[ -f "$master_pid_file" ]]; then
     local pid_from_file
     pid_from_file=$(cat "$master_pid_file" 2>/dev/null || true)
     if [[ "$pid_from_file" =~ ^[0-9]+$ ]]; then
       if kill -0 "$pid_from_file" 2>/dev/null; then
         old_pids="$pid_from_file"
-        log_info "从 PID 文件检测到旧 master 实例（PID: $old_pids）"
+        log_info "Old master instance detected from PID file (PID: $old_pids)"
       else
-        log_info "PID 文件存在但进程已不在，清理过期 PID 文件"
+        log_info "PID The file exists but the process is no longer there, clean up expired PID files"
         rm -f "$master_pid_file"
       fi
     fi
   fi
 
-  # 方案2：使用 pgrep 查找（支持多种运行方式）
+  # Option 2: Use pgrep to find (supports multiple running modes)
   if [[ -z "$old_pids" ]]; then
-    # 匹配模式：支持源码运行和 pipx 安装的方式
-    # - python.*master.py（源码运行）
-    # - Python.*master.py（macOS 上的 Python.app）
-    # - bot.py（pipx 安装的 master 别名）
+    # Matching mode: supports source code running and pipx installation methods
+    # - python.*master.py(Source code running)
+    # - Python.*master.py(macOS Python on.app)
+    # - bot.py(pipx installed master alias)
     local pgrep_pids
     pgrep_pids=$(pgrep -f "master\.py$" 2>/dev/null || true)
     if [[ -n "$pgrep_pids" ]]; then
       old_pids="$pgrep_pids"
-      log_info "通过 pgrep 检测到旧 master 实例（PID: $old_pids）"
+      log_info "Old master instance detected via pgrep (PID: $old_pids)"
     fi
   fi
 
-  # 如果两种方式都没找到，说明没有旧进程
+  # If neither method is found, it means there is no old process.
   if [[ -z "$old_pids" ]]; then
-    log_info "未检测到旧 master 实例"
+    log_info "Old master instance not detected"
     return 0
   fi
 
-  # 开始清理旧进程
-  log_info "正在优雅终止旧 master 实例（PID: $old_pids）..."
+  # Start cleaning up old processes
+  log_info "Gracefully terminating old master instance (PID: $old_pids)..."
 
-  # 发送 SIGTERM 信号优雅终止
+  # Send SIGTERM signal to terminate gracefully
   for pid in $old_pids; do
     kill -15 "$pid" 2>/dev/null || true
   done
 
-  # 循环等待进程退出
+  # Loop waiting for process to exit
   while (( waited < max_wait )); do
     sleep 1
     ((waited++))
 
-    # 检查所有 PID 是否都已退出
+    # Check if all PIDs have exited
     local all_exited=1
     for pid in $old_pids; do
       if kill -0 "$pid" 2>/dev/null; then
@@ -265,20 +276,20 @@ cleanup_old_master() {
     done
 
     if (( all_exited )); then
-      log_info "✅ 旧 master 已优雅退出（耗时 ${waited}秒）"
+      log_info "OK: The old master exited gracefully (elapsed ${waited}s)"
       rm -f "$master_pid_file"
       return 0
     fi
   done
 
-  # 优雅终止超时，执行强制结束
-  log_info "优雅终止超时（${max_wait}秒），执行强制结束..."
+  # Graceful termination timeout, execution forced end
+  log_info "graceful termination timeout (${max_wait}Second), Execute forced end..."
   for pid in $old_pids; do
     kill -9 "$pid" 2>/dev/null || true
   done
   sleep 2
 
-  # 最后检查
+  # final check
   local remaining_pids=""
   for pid in $old_pids; do
     if kill -0 "$pid" 2>/dev/null; then
@@ -287,75 +298,91 @@ cleanup_old_master() {
   done
 
   if [[ -n "$remaining_pids" ]]; then
-    log_error "❌ 无法清理旧 master 进程（残留 PID:$remaining_pids）"
-    log_error "请手动执行: kill -9$remaining_pids"
+    log_error "ERROR: Unable to clean up old master process (residual PID:$remaining_pids)"
+    log_error "Please execute manually: kill -9$remaining_pids"
     exit 1
   fi
 
-  log_info "✅ 旧 master 实例已强制清理"
+  log_info "OK: The old master instance has been forcefully cleaned"
   rm -f "$master_pid_file"
   return 0
 }
 
-# 调用清理函数
+# Call the cleanup function
 cleanup_old_master
 
-# 智能依赖管理：仅在必要时安装
+# Smart dependency management: install only when necessary
 REQUIREMENTS_FILE="${VIBEGO_REQUIREMENTS_PATH:-$ROOT_DIR/scripts/requirements.txt}"
 if [[ ! -f "$REQUIREMENTS_FILE" ]]; then
-  log_error "依赖文件缺失: $REQUIREMENTS_FILE"
+  log_error "Dependency files are missing: $REQUIREMENTS_FILE"
   exit 1
 fi
 
 PYTHON_BIN="$(select_python_binary)"
 
-# 检查是否需要安装依赖
+# Compatible with oldVersion: If it is detected in the warehouse .venv, then migrate to the runtime directory
+migrate_legacy_venv() {
+  if [[ -d "$LEGACY_VENV_DIR" && ! -e "$VENV_DIR" ]]; then
+    log_info "Old virtual environment directory detected:$LEGACY_VENV_DIR, Prepare to migrate to $VENV_DIR"
+    if mv "$LEGACY_VENV_DIR" "$VENV_DIR"; then
+      log_info "The virtual environment has been migrated to:$VENV_DIR"
+    else
+      log_error "Migration of the old virtual environment failed. Please check manually and try again."
+    fi
+  fi
+}
+
+migrate_legacy_venv
+
+# Check if dependencies need to be installed
 if check_deps_installed; then
-  log_info "依赖已安装且完整，跳过pip install（加速重启）"
-  source .venv/bin/activate
+  log_info "Dependencies are installed and complete, jump overpip install(accelerated restart)"
+  # shellcheck disable=SC1091
+  source "$VENV_DIR/bin/activate"
 else
-  log_info "首次启动或依赖缺失，正在安装依赖..."
+  log_info "First startup or missing dependencies, installing dependencies..."
 
-  # 创建或重建虚拟环境
-  "$PYTHON_BIN" -m venv .venv
-  source .venv/bin/activate
+  # Create or rebuild a virtual environment
+  "$PYTHON_BIN" -m venv "$VENV_DIR"
+  # shellcheck disable=SC1091
+  source "$VENV_DIR/bin/activate"
 
-  # 安装依赖
-  # 将 pip 输出重定向到日志文件，避免 BrokenPipe 错误
-  log_info "开始执行 pip install -r $REQUIREMENTS_FILE"
+  # Install dependencies
+  # Redirect pip output to log file to avoid BrokenPipe errors
+  log_info "Start executing pip install -r $REQUIREMENTS_FILE"
   PIP_LOG_FILE="$LOG_DIR/pip_install_$(date +%Y%m%d_%H%M%S).log"
   if pip install -r "$REQUIREMENTS_FILE" > "$PIP_LOG_FILE" 2>&1; then
-    log_info "依赖安装完成"
+    log_info "Dependency installation completed"
   else
     PIP_EXIT_CODE=$?
-    log_error "pip install 失败，退出码=$PIP_EXIT_CODE，详见 $PIP_LOG_FILE"
-    # 如果是 BrokenPipe (退出码 141)，验证依赖是否实际已安装
+    log_error "pip install Failure, exit code=$PIP_EXIT_CODE, See details $PIP_LOG_FILE"
+    # If it is BrokenPipe (Exit code 141), Verify that dependencies are actually installed
     if [[ $PIP_EXIT_CODE -eq 141 ]]; then
-      log_info "检测到 BrokenPipe 错误，验证依赖完整性..."
-      # 验证关键依赖包是否可导入
-      if "$ROOT_DIR/.venv/bin/python" -c "import aiogram, aiohttp, aiosqlite" 2>/dev/null; then
-        log_info "依赖验证通过，BrokenPipe 可忽略，继续执行"
+      log_info "BrokenPipe error detected, dependency integrity verified..."
+      # Verify whether key dependency packages can be imported
+      if "$VENV_DIR/bin/python" -c "import aiogram, aiohttp, aiosqlite" 2>/dev/null; then
+        log_info "Dependency verification passed, BrokenPipe can be ignored and execution continues."
       else
-        log_error "依赖验证失败，虽然是 BrokenPipe 但依赖未完整安装"
+        log_error "Dependency verification failed. Although it is BrokenPipe, the dependency is not completely installed."
         exit 1
       fi
     else
-      # 其他错误直接退出
+      # If there are other errors, exit directly.
       exit $PIP_EXIT_CODE
     fi
   fi
 fi
 
-# 后台启动 master，日志落在 vibe.log
-# 显式传递重启标记环境变量（如果存在）
+# Start the master in the background, and the log falls in vibe.log
+# Explicitly pass the restart flag environment variable if present
 if [[ -n "${MASTER_RESTART_EXPECTED:-}" ]]; then
-  log_info "检测到重启标记环境变量 MASTER_RESTART_EXPECTED=$MASTER_RESTART_EXPECTED"
+  log_info "Restart flag environment variable MASTER detected_RESTART_EXPECTED=$MASTER_RESTART_EXPECTED"
   export MASTER_RESTART_EXPECTED
 fi
 
-log_info "准备启动 master 进程..."
+log_info "Prepare to start the master process..."
 
-# 清理旧的错误日志（保留最近 10 次）
+# Clean old error logs (keep the last 10)
 cleanup_old_error_logs() {
   local error_log_pattern="$LOG_DIR/master_error_*.log"
   local error_logs
@@ -366,7 +393,7 @@ cleanup_old_error_logs() {
       ((count++))
       if (( count > 10 )); then
         rm -f "$logfile"
-        log_info "已清理旧错误日志: $logfile"
+        log_info "Cleaned old error logs: $logfile"
       fi
     done <<< "$error_logs"
   fi
@@ -374,73 +401,73 @@ cleanup_old_error_logs() {
 
 cleanup_old_error_logs
 
-# 创建带时间戳的错误日志文件
+# Create a timestamped error log file
 MASTER_ERROR_LOG="$LOG_DIR/master_error_$(date +%Y%m%d_%H%M%S).log"
 MASTER_STDOUT_LOG="$LOG_DIR/master_stdout.log"
 
-# 显式传递环境变量给 nohup 进程，确保重启信号文件路径正确
-# 使用虚拟环境的 Python 解释器，避免版本不匹配导致依赖加载失败
-# 重要：将 stderr 保存到日志文件，方便排查启动失败问题
-MASTER_RESTART_SIGNAL_PATH="$MASTER_RESTART_SIGNAL_PATH" nohup "$ROOT_DIR/.venv/bin/python" master.py > "$MASTER_STDOUT_LOG" 2> "$MASTER_ERROR_LOG" &
+# Explicitly pass environment variables to the nohup process to ensure that the restart signal file path is correct
+# Using virtual environment Python interpreter, avoidVersionMismatch causes dependency loading to fail
+# Important: Save stderr to a log file to facilitate troubleshooting startup failures.
+MASTER_RESTART_SIGNAL_PATH="$MASTER_RESTART_SIGNAL_PATH" nohup "$VENV_DIR/bin/python" master.py > "$MASTER_STDOUT_LOG" 2> "$MASTER_ERROR_LOG" &
 MASTER_PID=$!
 
-# 健壮性检查：确保进程成功启动
+# Robustness check: ensure processsuccessstart up
 if [[ -z "${MASTER_PID:-}" ]]; then
-  log_error "❌ 无法获取 master 进程 PID，启动失败"
-  log_error "可能原因：python 命令不可用或 master.py 有语法错误"
+  log_error "ERROR: Unable to obtain master process PID; startup failed"
+  log_error "Possible reasons: python command is not available or master.py There are grammatical errors"
   exit 1
 fi
 
-# 短暂等待后检查进程是否仍在运行
+# Check if the process is still running after a short wait
 sleep 0.5
 if ! kill -0 "$MASTER_PID" 2>/dev/null; then
-  log_error "❌ master 进程启动后立即退出（PID=$MASTER_PID）"
-  log_error "请检查："
-  log_error "  - master.py 是否有语法错误: python master.py"
-  log_error "  - 依赖是否完整: pip list | grep aiogram"
-  log_error "  - 错误日志: $MASTER_ERROR_LOG"
+  log_error "ERROR: Master process exited immediately after starting (PID=$MASTER_PID)"
+  log_error "Check, please:"
+  log_error "  - master.py whetherThere are grammatical errors: python master.py"
+  log_error "  - Are dependencies complete?: pip list | grep aiogram"
+  log_error "  - error log: $MASTER_ERROR_LOG"
 
-  # 输出错误日志的最后 20 行，帮助快速定位问题
+  # outputerror logthe end 20 OK, Help locate problems quickly
   if [[ -s "$MASTER_ERROR_LOG" ]]; then
     log_error ""
-    log_error "=== 错误日志最后 20 行 ==="
+    log_error "=== error logat last 20 OK ==="
     tail -20 "$MASTER_ERROR_LOG" | while IFS= read -r line; do
       log_error "  $line"
     done
     log_error "=========================="
   else
-    log_error "错误日志文件为空，可能是环境变量或路径问题"
+    log_error "error logFile is empty, It may be an environment variable or path problem"
   fi
 
   exit 1
 fi
 
-log_info "master 已后台启动，PID=$MASTER_PID，日志写入 ${LOG_FILE}"
+log_info "master Started in background, PID=$MASTER_PID, Log writing ${LOG_FILE}"
 
-# 健康检查：等待 master 上线并验证关键 worker
-log_info "开始执行健康检查..."
+# Health check: wait for master to come online and verify key workers
+log_info "Start master readiness check..."
 HEALTHCHECK_START=$(date +%s)
 
-if python scripts/master_healthcheck.py --project hyphavibebotbackend; then
+if python scripts/master_healthcheck.py --master-log "$LOG_FILE"; then
   HEALTHCHECK_END=$(date +%s)
   HEALTHCHECK_DURATION=$((HEALTHCHECK_END - HEALTHCHECK_START))
-  log_info "✅ master 健康检查通过，耗时 ${HEALTHCHECK_DURATION}s"
+  log_info "OK: Master readiness confirmed (elapsed ${HEALTHCHECK_DURATION}s)"
 else
   HEALTHCHECK_END=$(date +%s)
   HEALTHCHECK_DURATION=$((HEALTHCHECK_END - HEALTHCHECK_START))
-  log_error "⚠️ master 健康检查失败，耗时 ${HEALTHCHECK_DURATION}s"
-  log_error "建议检查："
-  log_error "  - 进程状态: ps aux | grep 'python.*master.py'"
-  log_error "  - 启动日志: tail -100 $LOG_DIR/start.log"
-  log_error "  - 运行日志: tail -100 $LOG_FILE"
-  log_error "  - 进程 PID: $MASTER_PID"
-
-  # 检查进程是否仍在运行
+  log_error "⚠️ master Health check failed, time consuming ${HEALTHCHECK_DURATION}s"
+  log_error "It is recommended to check:"
+  log_error "  - process status: ps aux | grep 'python.*master.py'"
+  log_error "  - Startup log: tail -100 $LOG_DIR/start.log"
+  log_error "  - Run log: tail -100 $LOG_FILE"
+  log_error "  - Process PID: $MASTER_PID"
+
+  # Check if the process is still running
   if kill -0 "$MASTER_PID" 2>/dev/null; then
-    log_info "master 进程仍在运行（PID=$MASTER_PID），允许继续启动"
-    log_info "⚠️ 请手动验证服务是否正常工作"
+    log_info "master The process is still running (PID=$MASTER_PID), Allow startup to continue"
+    log_info "⚠️ Please manually verify that the service is working properly"
   else
-    log_error "❌ master 进程已退出，启动失败"
+    log_error "ERROR: Master process exited and failed to start"
     exit 1
   fi
 fi

scripts/start_tmux_codex.sh

@@ -18,8 +18,13 @@ MODEL_WORKDIR="${MODEL_WORKDIR:-$ROOT_DIR}"
 MODEL_SESSION_ROOT="${MODEL_SESSION_ROOT:-${CODEX_SESSION_ROOT:-$HOME/.codex/sessions}}"
 MODEL_SESSION_GLOB="${MODEL_SESSION_GLOB:-rollout-*.jsonl}"
 SESSION_POINTER_FILE="${SESSION_POINTER_FILE:-$LOG_ROOT/${MODEL_NAME:-codex}/${PROJECT_NAME:-project}/current_session.txt}"
+SESSION_LOCK_FILE="${SESSION_LOCK_FILE:-${SESSION_POINTER_FILE%.txt}.lock.json}"
+SESSION_CAPTURE_TIMEOUT="${SESSION_CAPTURE_TIMEOUT:-180}"
+SESSION_CAPTURE_POLL_INTERVAL="${SESSION_CAPTURE_POLL_INTERVAL:-0.5}"
+# Ensure optional CODEX session root is always defined even under `set -u`
+: "${CODEX_SESSIONS_ROOT:=}"
 
-# 避免 oh-my-zsh 在非交互环境弹出更新提示
+# Avoid oh-my-zsh popping up update prompts in non-interactive environments
 export DISABLE_UPDATE_PROMPT="${DISABLE_UPDATE_PROMPT:-true}"
 
 expand_path() {
@@ -40,7 +45,7 @@ KILL_SESSION=0
 
 usage() {
   cat <<USAGE
-用法：${0##*/} [--dry-run] [--force] [--restart] [--kill]
+usage:${0##*/} [--dry-run] [--force] [--restart] [--kill]
 USAGE
 }
 
@@ -51,13 +56,13 @@ while [[ $# -gt 0 ]]; do
     --restart) RESTART=1; FORCE_START=1 ;;
     --kill) KILL_SESSION=1 ;;
     -h|--help) usage; exit 0 ;;
-    *) echo "未知参数: $1" >&2; usage; exit 1 ;;
+    *) echo "unknown parameters: $1" >&2; usage; exit 1 ;;
   esac
   shift
 done
 
 if ! command -v tmux >/dev/null 2>&1; then
-  echo "tmux 未安装" >&2
+  echo "tmux Not installed" >&2
   exit 1
 fi
 
@@ -67,6 +72,7 @@ MODEL_SESSION_ROOT=$(expand_path "$MODEL_SESSION_ROOT")
 SESSION_POINTER_FILE=$(expand_path "$SESSION_POINTER_FILE")
 ensure_dir "$(dirname "$LOG_PATH")"
 ensure_dir "$(dirname "$SESSION_POINTER_FILE")"
+ensure_dir "$(dirname "$SESSION_LOCK_FILE")"
 
 run_tmux() {
   if (( DRY_RUN )); then
@@ -98,12 +104,12 @@ else
   fi
 fi
 
-# 启动前先进行一次清理，避免旧日志超限
+# Clean up before starting to avoid old logs exceeding the limit.
 if ! env \
   MODEL_LOG_MAX_BYTES="$MODEL_LOG_MAX_BYTES" \
   MODEL_LOG_RETENTION_SECONDS="$MODEL_LOG_RETENTION_SECONDS" \
   "$PYTHON_EXEC" "$LOG_WRITER" "$LOG_PATH" </dev/null; then
-  echo "预处理日志文件失败" >&2
+  echo "Preprocessing log file failed" >&2
   exit 1
 fi
 
@@ -115,7 +121,7 @@ printf -v PIPE_CMD 'env MODEL_LOG_MAX_BYTES=%q MODEL_LOG_RETENTION_SECONDS=%q %q
   "$LOG_PATH"
 run_tmux pipe-pane -o -t "$SESSION_NAME" "$PIPE_CMD"
 
-# 同步环境变量到 tmux 服务端，避免复用旧会话时丢失设置
+# Synchronize environment variables to the tmux server to avoid losing settings when reusing old sessions
 run_tmux set-environment -t "$SESSION_NAME" DISABLE_UPDATE_PROMPT "${DISABLE_UPDATE_PROMPT:-true}"
 if [[ -n "${CLAUDE_CODE_DISABLE_FILE_CHECKPOINTING:-}" ]]; then
   run_tmux set-environment -t "$SESSION_NAME" CLAUDE_CODE_DISABLE_FILE_CHECKPOINTING "${CLAUDE_CODE_DISABLE_FILE_CHECKPOINTING}"
@@ -138,10 +144,29 @@ fi
 
 
 if (( DRY_RUN )); then
-  printf '[dry-run] 会话日志路径: %s\n' "$SESSION_POINTER_FILE"
+  printf '[dry-run] Session log path: %s\n' "$SESSION_POINTER_FILE"
   exit 0
 fi
 
 : > "$SESSION_POINTER_FILE"
+rm -f "$SESSION_LOCK_FILE"
+
+if (( ! DRY_RUN )); then
+  WATCH_ARGS=("$PYTHON_EXEC" "$ROOT_DIR/scripts/session_pointer_watch.py" "--pointer" "$SESSION_POINTER_FILE" "--lock" "$SESSION_LOCK_FILE" "--glob" "$MODEL_SESSION_GLOB" "--workdir" "$MODEL_WORKDIR" "--tmux-session" "$SESSION_NAME" "--project" "${PROJECT_NAME:-}")
+  if [[ -n "$MODEL_SESSION_ROOT" ]]; then
+    WATCH_ARGS+=("--session-root" "$MODEL_SESSION_ROOT")
+  fi
+  if [[ -n "$CODEX_SESSIONS_ROOT" ]]; then
+    WATCH_ARGS+=("--additional-root" "$CODEX_SESSIONS_ROOT")
+  fi
+  WATCH_ARGS+=("--additional-root" "$(dirname "$SESSION_POINTER_FILE")")
+  WATCH_ARGS+=("--additional-root" "$(dirname "$SESSION_POINTER_FILE")/sessions")
+  WATCH_ARGS+=("--timeout" "$SESSION_CAPTURE_TIMEOUT")
+  WATCH_ARGS+=("--poll" "$SESSION_CAPTURE_POLL_INTERVAL")
+  if ! "${WATCH_ARGS[@]}"; then
+    echo "Failed to capture Codex/Claude session log for tmux session '$SESSION_NAME'. Please check CLI startup output." >&2
+    exit 1
+  fi
+fi
 
 exit 0