ATH-MaaS/Pixelle-Video · error · RuntimeError

万象视频任务查询失败: status={rsp.status_code}, code={rsp.code}, messa

Error message

万象视频任务查询失败: status={rsp.status_code}, code={rsp.code}, message={rsp.message}, task_id={task_id}

What it means

For async submissions, the client polls VideoSynthesis.wait() (wrapped with retries, max_attempts=8, base_delay=5s). If the final polled response has a non-OK status_code, this RuntimeError is raised with the HTTP status, API code/message, and the task_id so the failed task can be traced in the DashScope console.

Source

Thrown at pixelle_video/services/api_services/video_dashscope.py:383

        if not video_url:
            task_id = self._extract_task_id(rsp)
            task_status = self._extract_task_status(rsp)
            if not task_id:
                raise RuntimeError(
                    "万象视频 API 未返回 video_url 或 task_id，无法查询结果: "
                    f"status={rsp.status_code}, code={rsp.code}, message={rsp.message}, "
                    f"task_status={task_status}"
                )

            logger.info(f"DashscopeVideoClient: 任务已提交 task_id={task_id}, status={task_status}; 等待生成完成...")
            rsp = self._with_network_retry(
                f"wait task {task_id}",
                lambda: VideoSynthesis.wait(task=rsp, api_key=self.api_key),
                max_attempts=8,
                base_delay=5.0,
            )
            if rsp.status_code != HTTPStatus.OK:
                raise RuntimeError(
                    f"万象视频任务查询失败: status={rsp.status_code}, "
                    f"code={rsp.code}, message={rsp.message}, task_id={task_id}"
                )

            video_url = self._extract_video_url(rsp)
            task_status = self._extract_task_status(rsp)
            if not video_url:
                raise RuntimeError(
                    "万象视频任务完成后仍未返回 video_url: "
                    f"code={rsp.code}, message={rsp.message}, task_id={task_id}, task_status={task_status}, "
                    f"output={self._safe_output_repr(rsp)}"
                )

        logger.info(f"DashscopeVideoClient: 视频生成成功: {video_url}")

        # 确保输出目录存在
        os.makedirs(os.path.dirname(save_path), exist_ok=True)

View on GitHub (pinned to 848b054e4f)

Solutions

Use the task_id from the error to query the task status/details directly in the DashScope console or via the query API.
Read the embedded code/message for the concrete cause (auth vs task-failed vs throttling).
Re-submit the task if it failed terminally; check moderation/billing causes first.
Increase polling tolerance (max_attempts/base_delay) for long-running models if timeouts are the cause.
Verify the same valid api_key is used for submit and wait calls.

Example fix

// before
video = client.generate_video(...)  # RuntimeError on wait
// after
try:
    video = client.generate_video(...)
except RuntimeError as e:
    task_id = extract_task_id_from(str(e))  # from error text
    logger.error("task %s failed, check dashscope console", task_id)
    raise

Defensive patterns

Strategy: retry

Validate before calling

# before submitting long tasks, confirm API key and quota are healthy
if not os.environ.get("DASHSCOPE_API_KEY"):
    raise ValueError("DASHSCOPE_API_KEY not set")

Type guard

def wait_rsp_ok(rsp) -> bool:
    return getattr(rsp, "status_code", None) == 200

Try / catch

try:
    video = client.generate_video(...)
except RuntimeError as e:
    # extract task_id from message and query console/API for final status
    logger.error("dashscope task wait failed: %s", e)
    raise

Prevention

Record task_id from logs at submit time for later diagnosis.
Use the same, valid api_key for submit and polling.
Allow generous polling windows for long renders; the client waits max_attempts=8 with 5s base delay.
Investigate embedded code/message to distinguish terminal task failure from transient poll errors.

When it happens

Trigger: Polling task {task_id} ends with a non-200 response: query-auth failure, task record expired/unknown, service error during wait, or quota/billing rejection surfacing at poll time.

Common situations: Task failed server-side (rendering error, content moderation) and the wait response reflects it; API key rotated between submit and wait; task submitted long ago so the record aged out; transient DashScope outages exceeding the 8 retry attempts.

Related errors

AI-assisted analysis of ATH-MaaS/Pixelle-Video@848b054e4f (2026-08-30). Data as JSON: /api/errors/c1c198634e65750a. Report an issue: GitHub.