PlantAI-rag-new/agent/qa/scripts/start.sh

#!/usr/bin/env bash
# agent/qa/scripts/start.sh —— QA V1.1.73 服务唯一启动接口
# 用法：bash scripts/start.sh [port]   （默认 8091 终态；浸泡期影子用 7921）
set -euo pipefail
AGENT_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
PORT="${1:-8091}"  # 终态端口
ENV_FILE="${QA_ENV_FILE:-$AGENT_ROOT/env/prod.env}"
LOG_FILE="${QA_LOG_FILE:-/tmp/plantai-agent-qa-$PORT.log}"
PID_FILE="$AGENT_ROOT/run/qa-$PORT.pid"
PYTHON="${QA_PYTHON:-$AGENT_ROOT/../venv/bin/python}"
[[ -x "$PYTHON" ]] || PYTHON="/home/xzq/5090/teamWorkspace/zh/plantai-rag-new-venv/bin/python"

mkdir -p "$AGENT_ROOT/run"
[[ -x "$PYTHON" ]] || { echo "QA Python 不可执行: $PYTHON" >&2; exit 1; }
[[ -f "$ENV_FILE" ]] || { echo "缺少 env 文件: $ENV_FILE" >&2; exit 1; }

if [[ -f "$PID_FILE" ]] && kill -0 "$(cat "$PID_FILE")" 2>/dev/null; then
  echo "QA 已在运行 pid=$(cat "$PID_FILE")" >&2
  exit 1
fi

set -a
# shellcheck disable=SC1090
. "$ENV_FILE"
set +a

if [[ -f "$AGENT_ROOT/env/active_library.env" ]]; then
  set -a; . "$AGENT_ROOT/env/active_library.env"; set +a
fi


# ---- 开机自愈:后端校验(Qdrant/Neo4j)失败即崩且零重试,启动前必须等依赖就绪(2026-09-23) ----
QD_READYZ="${QDRANT_URL:-http://127.0.0.1:27433}/readyz"
QD_DEADLINE=$(( $(date +%s) + 1800 ))
until curl -sf --max-time 3 "$QD_READYZ" >/dev/null 2>&1; do
  [ "$(date +%s)" -ge "$QD_DEADLINE" ] && { echo "[qa] Qdrant 30 分钟未就绪,放弃启动" >&2; exit 1; }
  sleep 5
done
N4_HP="${NEO4J_URI#bolt://}"; N4_H="${N4_HP%:*}"; N4_P="${N4_HP##*:}"
N4_DEADLINE=$(( $(date +%s) + 600 ))
until timeout 3 bash -c "echo > /dev/tcp/$N4_H/$N4_P" 2>/dev/null; do
  [ "$(date +%s)" -ge "$N4_DEADLINE" ] && { echo "[qa] Neo4j 10 分钟未就绪,放弃启动" >&2; exit 1; }
  sleep 5
done
# ---- 就绪等待结束 ----

# reranker 模型路径回退：env 未指定时用安装包内模型（install/models/bge-reranker-base）
if [[ -z "${QA_V1_BGE_RERANKER_MODEL:-}" && -d "$AGENT_ROOT/../models/bge-reranker-base" ]]; then
  export QA_V1_BGE_RERANKER_MODEL="$AGENT_ROOT/../models/bge-reranker-base"
fi

cd "$AGENT_ROOT"
setsid "$PYTHON" -m uvicorn scripts.qa_v1_1_73_server:app \
  --host 127.0.0.1 --port "$PORT" \
  > "$LOG_FILE" 2>&1 < /dev/null &
echo $! > "$PID_FILE"
disown

for _ in $(seq 1 30); do
  if curl -sf --max-time 3 "http://127.0.0.1:$PORT/health" >/dev/null 2>&1; then
    echo "QA UP (127.0.0.1:$PORT, pid=$(cat "$PID_FILE"))"
    exit 0
  fi
  sleep 2
done
echo "QA FAIL —— 日志尾部：" >&2
tail -30 "$LOG_FILE" >&2
exit 1