fix：排除资源备份表

scripts/resource-restore/apply-resource-backup.sh

@@ -32,6 +32,8 @@ MQ_PLAN="${WORK_DIR}/restore-mq.tsv"
 ENV_FILE="${WORK_DIR}/restore-env.sh"
 STOP_COMMANDS_PLAN="${WORK_DIR}/restore-stop-commands.tsv"
 START_COMMANDS_PLAN="${WORK_DIR}/restore-start-commands.tsv"
+SERVICES_STOPPED_FOR_RESTORE=false
+SERVICES_STARTING_AFTER_FAILURE=false
 
 json_escape() {
   local value="$1"
@@ -63,6 +65,7 @@ fail() {
   local phase="${2:-FAILED}"
   log "[ERROR] ${message}"
   write_status "FAILED" "${phase}" "${message}"
+  attempt_service_start_after_failure "${phase}"
   exit 1
 }
 
@@ -101,6 +104,9 @@ restore_databases() {
   require_env "DB_USER" "DATABASE"
   local mysql_path="${MYSQL_PATH:-mysql}"
   local db_password="${DB_PASSWORD:-}"
+  local timeout_seconds="${RESTORE_DB_TIMEOUT_SECONDS:-300}"
+  [[ "${timeout_seconds}" =~ ^[0-9]+$ ]] || timeout_seconds=300
+  (( timeout_seconds > 0 )) || timeout_seconds=300
   while IFS=$'\t' read -r entry_path db_name || [[ -n "${entry_path:-}" ]]; do
     [[ -n "${entry_path:-}" ]] || continue
     [[ -n "${db_name:-}" ]] || fail "database name is empty for ${entry_path}" "DATABASE"
@@ -113,7 +119,32 @@ restore_databases() {
       --username "${DB_USER}" \
       --password "${db_password}" \
       --db-name "${db_name}" \
-      --sql-file "${sql_file}" >> "${LOG_FILE}" 2>&1
+      --sql-file "${sql_file}" >> "${LOG_FILE}" 2>&1 &
+    local db_pid=$!
+    local timeout_marker="${WORK_DIR}/.db-restore-timeout-${db_pid}"
+    rm -f "${timeout_marker}" || true
+    (
+      sleep "${timeout_seconds}"
+      if kill -0 "${db_pid}" >/dev/null 2>&1; then
+        printf 'timeout' > "${timeout_marker}"
+        kill "${db_pid}" >/dev/null 2>&1 || true
+        sleep 1
+        kill -9 "${db_pid}" >/dev/null 2>&1 || true
+      fi
+    ) &
+    local watchdog_pid=$!
+    local db_exit=0
+    wait "${db_pid}" || db_exit=$?
+    kill "${watchdog_pid}" >/dev/null 2>&1 || true
+    wait "${watchdog_pid}" >/dev/null 2>&1 || true
+    if [[ -f "${timeout_marker}" ]]; then
+      rm -f "${timeout_marker}" || true
+      fail "database restore timed out after ${timeout_seconds}s: ${db_name}" "DATABASE"
+    fi
+    rm -f "${timeout_marker}" || true
+    if [[ "${db_exit}" -ne 0 ]]; then
+      fail "database restore failed: ${db_name}" "DATABASE"
+    fi
     log "restored database: ${db_name}"
   done < "${DATABASES_PLAN}"
 }
@@ -153,6 +184,49 @@ run_service_commands() {
   return 0
 }
 
+run_service_commands_best_effort() {
+  local plan_file="$1"
+  local phase="$2"
+  [[ -f "${plan_file}" ]] || return 1
+  [[ -s "${plan_file}" ]] || return 1
+  while IFS=$'\t' read -r code required command || [[ -n "${code:-}" ]]; do
+    [[ -n "${code:-}" ]] || continue
+    if [[ -z "${command:-}" ]]; then
+      log "[WARN] ${phase} ${code}: command is empty"
+      continue
+    fi
+    log "${phase} ${code}: ${command}"
+    if bash -lc "${command}" >> "${LOG_FILE}" 2>&1; then
+      log "${phase} ${code}: success"
+    else
+      log "[WARN] ${phase} ${code}: failed while recovering from restore failure"
+    fi
+  done < "${plan_file}"
+  return 0
+}
+
+attempt_service_start_after_failure() {
+  local failed_phase="${1:-FAILED}"
+  if [[ "${SERVICES_STOPPED_FOR_RESTORE}" != "true" ]]; then
+    return 0
+  fi
+  if [[ "${SERVICES_STARTING_AFTER_FAILURE}" == "true" || "${failed_phase}" == "SERVICE_START" ]]; then
+    return 0
+  fi
+  SERVICES_STARTING_AFTER_FAILURE=true
+  log "SERVICE_START_AFTER_FAILURE due to ${failed_phase}"
+  if run_service_commands_best_effort "${START_COMMANDS_PLAN}" "SERVICE_START_AFTER_FAILURE"; then
+    SERVICES_STOPPED_FOR_RESTORE=false
+    return 0
+  fi
+  if [[ -n "${TMS_SCRIPT_PATH:-}" && -x "${TMS_SCRIPT_PATH}" ]]; then
+    "${TMS_SCRIPT_PATH}" start >> "${LOG_FILE}" 2>&1 \
+      && log "SERVICE_START_AFTER_FAILURE TMS fallback: success" \
+      || log "[WARN] SERVICE_START_AFTER_FAILURE TMS fallback: failed"
+    SERVICES_STOPPED_FOR_RESTORE=false
+  fi
+}
+
 health_check() {
   local url="${HEALTH_CHECK_URL:-}"
   [[ -n "${url}" ]] || {
@@ -217,6 +291,7 @@ fi
 # 因为某些部署环境可能由 systemd/外部平台接管服务状态，但文件/数据库/MQ 失败必须中断。
 write_status "RUNNING" "SERVICE_STOP" "stopping services"
 log "SERVICE_STOP"
+SERVICES_STOPPED_FOR_RESTORE=true
 if ! run_service_commands "${STOP_COMMANDS_PLAN}" "SERVICE_STOP" \
     && [[ -n "${TMS_SCRIPT_PATH:-}" && -x "${TMS_SCRIPT_PATH}" ]]; then
   "${TMS_SCRIPT_PATH}" stop >> "${LOG_FILE}" 2>&1 || true
@@ -240,6 +315,7 @@ if ! run_service_commands "${START_COMMANDS_PLAN}" "SERVICE_START" \
     && [[ -n "${TMS_SCRIPT_PATH:-}" && -x "${TMS_SCRIPT_PATH}" ]]; then
   "${TMS_SCRIPT_PATH}" start >> "${LOG_FILE}" 2>&1 || true
 fi
+SERVICES_STOPPED_FOR_RESTORE=false
 
 write_status "RUNNING" "HEALTH_CHECK" "checking health"
 log "HEALTH_CHECK"