666ghj/MiroFish · error · ValueError

模拟环境未运行或已关闭，无法执行Interview

Error message

模拟环境未运行或已关闭，无法执行Interview: {simulation_id}

What it means

After confirming the run directory exists, interview_agent builds a SimulationIPCClient over the run's IPC channel and calls check_env_alive(); if the OASIS environment process inside the simulation is not alive (not started, exited, or IPC socket dead), it refuses to send the Interview command with this ValueError.

Solutions

1. Retry with backoff for the first ~30-60s after start until check_env_alive() passes (env boot takes time).
2. Check the simulation's runner_status and env logs — if the env process exited, restart the simulation before interviewing.
3. If a stale IPC socket is suspected (crashed prior run), clean the run directory / restart the simulation.

Example fix

// before
resp = SimulationRunner.interview_agent(sim_id, agent_id, prompt);

// after
from app.services.simulation_ipc import SimulationIPCClient
client = SimulationIPCClient(sim_dir)
for _ in range(12):  # up to ~60s for env boot
    if client.check_env_alive():
        break
    time.sleep(5)
else:
    raise ServiceUnavailable(f"env for {sim_id} not alive")
resp = SimulationRunner.interview_agent(sim_id, agent_id, prompt);

from app.services.simulation_ipc import SimulationIPCClient

def env_ready(sim_id: str) -> bool:
    sim_dir = Path(SimulationRunner.RUN_STATE_DIR) / sim_id
    return sim_dir.is_dir() and SimulationIPCClient(str(sim_dir)).check_env_alive()

try:
    SimulationRunner.interview_agent(sim_id, agent_id, prompt)
except ValueError as e:
    if "环境未运行" in str(e):
        return {"success": False, "retry_after": 5}  # client backs off
    raise

When it happens

Trigger: Interviewing while the simulation is still STARTING (env process not yet listening), after the env process has exited/crashed, after the simulation was stopped, or when the IPC socket/pipe in sim_dir is stale from a previous run.

Common situations: Frontend enables the interview panel based on run existence rather than env liveness; polling too early after start; env crashed mid-run while the runner state still says RUNNING.

Related errors

• 等待命令响应超时 ( 秒)
• 模拟配置中没有Agent
• 没有找到符合条件的实体，请检查图谱是否正确构建
• 模拟停止失败
• 启用图谱记忆更新时必须提供 graph_id

AI-assisted analysis of 666ghj/MiroFish@b5b53acc57 (2026-08-14). Data as JSON: /api/errors/538768bf86a46230. Report an issue: GitHub.

Appendix: source

Thrown at backend/app/services/simulation_runner.py:1729

                - "reddit": 只采访Reddit平台
                - None: 双平台模拟时同时采访两个平台，返回整合结果
            timeout: 超时时间（秒）

        Returns:
            采访结果字典

        Raises:
            ValueError: 模拟不存在或环境未运行
            TimeoutError: 等待响应超时
        """
        sim_dir = os.path.join(cls.RUN_STATE_DIR, simulation_id)
        if not os.path.exists(sim_dir):
            raise ValueError(f"模拟不存在: {simulation_id}")

        ipc_client = SimulationIPCClient(sim_dir)

        if not ipc_client.check_env_alive():
            raise ValueError(f"模拟环境未运行或已关闭，无法执行Interview: {simulation_id}")

        logger.info(f"发送Interview命令: simulation_id={simulation_id}, agent_id={agent_id}, platform={platform}")

        response = ipc_client.send_interview(
            agent_id=agent_id,
            prompt=prompt,
            platform=platform,
            timeout=timeout
        )

        if response.status.value == "completed":
            return {
                "success": True,
                "agent_id": agent_id,
                "prompt": prompt,
                "result": response.result,
                "timestamp": response.timestamp
            }

View on GitHub (pinned to b5b53acc57)