一、软硬件信息
1.服务器厂家: 云 GPU
2.沐曦GPU型号：C500 16GB
3.操作系统内核版本：Ubuntu 22.04
4.是否开启CPU虚拟化：未知
5.mx-smi回显：
=======================================================
__ __ _ _ ___ ____ _ _
| \/ | ___ __ _ _ __| | __ / \ |_ _| / ___(_) |_ ___ ___
| |\/| |/ _ \ / _` | '__| |/ / / _ \ | | _____ | | _| | __/ _ \/ _ \
| | | | (_) | (_| | | | < / ___ \ | | |_____| | |_| | | || __/ __/
|_| |_|\___/ \__,_|_| |_|\_\ /_/ \_\___| \____|_|\__\___|\___|
========================================
Moark- Gitee-AI 欢迎您！
=======================================================
为了保证您的数据安全与使用体验，请您仔细阅读并严格遵守以下规则：
一、数据持久化存储
容器实例的存储机制非常特殊，我们强烈建议您将所有需要长期保存的数据、代码、模型等重要文件，全部存放在 `/root` 或 `/data` 目录下。
- 持久化目录：`/root` 和 `/data` 是唯二的持久化目录。在这两个目录下的数据，实例进行`更换镜像`或`重启`操作后**仍会保留**。
- 非持久化目录的数据丢失风险：
-- 更换镜像：当您更换实例的镜像时，只有 `/root` 和 `/data` 目录下的数据会被完整保留。所有其他目录（如 `/home`, `/tmp` 等）都将被重置为新镜像的初始状态，您存放在这些路径下的**数据将会永久丢失**。
-- 实例重启：在极少数情况下，执行“重启”操作也可能会导致 `/root` 和 `/data` 之外的其他目录数据丢失。
二、 访问容器内Web服务
出于安全考虑，容器实例暂不提供直接对外访问的公网端口。如需访问您在容器内部署的Web应用或API，您可以通过隧道代理的方式从本地主机安全连接并访问。
三、帮助文档与资源
- 关于如何上传/下载数据和隧道代理连接教程，请访问: https://ai.gitee.com/docs/compute/container
- 有关GPU使用状态查询（`mx-smi` 命令）及相关教程，请访问： https://ai.gitee.com/docs/compute/domestic-gpu
四、重要声明
- 请注意：ai.gitee 的所有算力资源严禁用于任何形式的加密货币挖矿活动。一经发现，我们将有权立即终止服务并追究相关责任。
=======================================================
=== GPU信息（mx-smi）===
mx-smi version: 2.2.12
=================== MetaX System Management Interface Log ===================
Timestamp : Sat May 30 16:32:13 2026
Attached GPUs : 1
+---------------------------------------------------------------------------------+
| MX-SMI 2.2.12 Kernel Mode Driver Version: 3.0.11 |
| MACA Version: 3.5.3.20 BIOS Version: 1.27.5.0 |
|------------------+-----------------+---------------------+----------------------|
| Board Name | GPU Persist-M | Bus-id | GPU-Util sGPU-M |
| Pwr:Usage/Cap | Temp Perf | Memory-Usage | GPU-State |
|==================+=================+=====================+======================|
| 0 MetaX C500 | 0 Off | 0000:10:00.0 | 0% Enabled |
| 52W / 350W | 35C P0 | 826/65536 MiB | Available |
+------------------+-----------------+---------------------+----------------------+
+---------------------------------------------------------------------------------+
| Sliced GPU |
|------------------------------------+---------------------+----------------------|
| Minor GPU sGPU-Id Compute | Vram Quota | sGPU-Util |
|====================================+=====================+======================|
| 004 0 0 25% | 0/16000 MiB | 0% |
+------------------------------------+---------------------+----------------------+
+---------------------------------------------------------------------------------+
| Process: |
| GPU PID Process Name GPU Memory |
| Usage(MiB) |
|=================================================================================|
| no process found |
+---------------------------------------------------------------------------------+
End of Log
6.docker info回显：
7.镜像版本：vLLM / 0.17.0 / Python 3.10 / maca 3.5.3.307
8.启动容器命令：未知
9.容器内执行命令：
二、问题现象
开启：profiler_server.exe、gui-profiler.exe。
gui-profiler.exe 已在图中填写好对应的字段
显示增加任务成功，但是没有数据生成。
profiler_server.exe 报错：
[info] start a new perf exec thread
[info] start perf_exec
[info] new exec 58bee9a7-829d-4eaa-9cb4-fd779b2d7bf9:('export MACA_PATH=/opt/maca/; export LD_LIBRARY_PATH=/opt/maca/lib/; /bin/hostname', 'maca-profiler-probe', ['WORKGROUPS', 'WAVES', 'Average Wave life cycles', 'instruction per cycle', 'instruction throughput', 'AP active cycles', 'AP MTE Duty ratio', 'AP STE Duty ratio', 'AP MMA Duty ratio', 'VLS Duty ratio', 'L2C Duty ratio', 'Compute Instructions Cycles Rate', 'DPC Computing cycles comparison', 'Instructions Comparison', 'instruction number', 'instructions per AP', 'Global Read Instructions', 'Global Write Instructions', 'Global Atomic Instructions', 'Private Read Instructions', 'Private Write Instructions', 'Constant Read Instructions', 'Constant Read SL1 Instructions', 'Constant Read L2 Instructions', 'Constant SL1 Hit Rate', 'VL1 Read Instructions', 'VL1 Write Instructions', 'VL1 Atomic Instructions', 'VL1 Hit Rate', 'L2C Read Instructions', 'L2C Write Instructions', 'L2C Atomic Instructions', 'L2C Hit Rate', 'Memory Data Flow Chart', 'Dnoc Read Average Latency', 'Dnoc Read Latency Histogram', 'VL1 partition stall cycles layout', 'Achieved waves', 'Dispatched waves', 'Total Instructions', 'Compute Instructions', 'Memory Instructions', 'Total Cycles', 'Memory Access per Second', 'Average AP busy cycles', 'AP busy Duty', 'Compute Instructions busy Duty', 'load instructions', 'store instructions', 'atomic instructions', 'total instructions', 'average latency per load instruction', 'average latency per store instruction', 'average latency per atomic instruction', 'average latency of all stages per instruction', 'average cycles per load instruction', 'average cycles per store instruction', 'average cycles per atomic instruction', 'average conflict cycles per instruction', 'average cycles per instruction', 'shared memory access efficiency'])
Exception in thread Thread-1705:
Traceback (most recent call last):
File "threading.py", line 917, in _bootstrap_inner
File "threading.py", line 865, in run
File "phttp\server_backend.py", line 686, in perf_exec
File "phttp\server_backend.py", line 228, in _get_profile_path
File "common\api_common.py", line 378, in remote_ret_code_ssh
File "paramiko\client.py", line 496, in connect
File "paramiko\client.py", line 818, in _auth
File "paramiko\client.py", line 805, in _auth
File "paramiko\transport.py", line 1587, in auth_password
File "paramiko\auth_handler.py", line 248, in wait_for_response
paramiko.ssh_exception.AuthenticationException: Authentication failed: transport shut down or saw EOF