"""重跑 4 个失败任务(v3),捕获完整错误输出""" import os, sys, subprocess, time from pathlib import Path from datetime import datetime from dotenv import load_dotenv from zoneinfo import ZoneInfo load_dotenv(Path(__file__).resolve().parents[2] / ".env") TZ = ZoneInfo("Asia/Shanghai") ETL_CWD = Path(__file__).resolve().parents[2] / "apps" / "etl" / "connectors" / "feiqiu" TASKS = [ "DWS_FINANCE_INCOME_STRUCTURE", "DWS_ASSISTANT_MONTHLY", "DWS_RELATION_INDEX", "DWS_SPENDING_POWER_INDEX", ] COMMON_ARGS = [ "--window-start", "2025-11-01 00:00:00", "--window-end", "2026-02-27 00:00:00", "--window-split-days", "30", "--force-full", ] def run_task(task: str) -> dict: layer = "INDEX" if task.endswith("_INDEX") else "DWS" cmd = [ "uv", "run", "--package", "etl-feiqiu", "python", "-m", "cli.main", "--layers", layer, "--tasks", task, *COMMON_ARGS, ] start = time.time() proc = subprocess.run( cmd, cwd=str(ETL_CWD), capture_output=True, text=True, encoding="utf-8", errors="replace", timeout=300, ) elapsed = time.time() - start output = proc.stdout + proc.stderr has_error = any( kw in line for line in output.splitlines() for kw in ("ERROR", "CRITICAL", "Traceback") ) return { "task": task, "exit_code": proc.returncode, "elapsed": elapsed, "success": proc.returncode == 0 and not has_error, "output": output, } def main(): now = datetime.now(TZ) print(f"{'='*60}") print(f"失败任务重跑 v3 | {now.isoformat()}") print(f"{'='*60}\n") results = [] for i, task in enumerate(TASKS, 1): print(f"[{i}/{len(TASKS)}] {task} ...", end=" ", flush=True) try: r = run_task(task) results.append(r) icon = "OK" if r["success"] else "FAIL" print(f"{icon} ({r['elapsed']:.0f}s)") if not r["success"]: # 打印最后 30 行 lines = r["output"].splitlines() for line in lines[-30:]: print(f" | {line}") except subprocess.TimeoutExpired: results.append({"task": task, "success": False, "elapsed": 300, "exit_code": -1, "output": "TIMEOUT"}) print("TIMEOUT") ok = [r for r in results if r["success"]] fail = [r for r in results if not r["success"]] print(f"\n{'='*60}") print(f"结果: {len(ok)}/{len(results)} 成功") # 保存日志 log_root = os.environ.get("SYSTEM_LOG_ROOT") if not log_root: raise RuntimeError("SYSTEM_LOG_ROOT 未设置") log_dir = Path(log_root) log_dir.mkdir(parents=True, exist_ok=True) log_file = log_dir / f"{now.strftime('%Y%m%d')}_rerun_v3.log" with open(log_file, "w", encoding="utf-8") as f: for r in results: icon = "OK" if r["success"] else "FAIL" f.write(f"[{icon}] {r['task']} ({r['elapsed']:.0f}s)\n") f.write(f"{r['output']}\n{'='*60}\n") print(f"日志: {log_file}") sys.exit(0 if not fail else 1) if __name__ == "__main__": main()