使用Azure ML Python SDK v2时Mlflow Run ID找不到，无法访问运行日志

使用Azure ML jobs通过Python SDK v2运行实验，运行完成后无法访问运行日志，系统提示“run 'xxxx' not found”。确认该Run ID确实存在，且本人是工作区和集群的所有者。

代码示例 #

from mlflow.tracking import MlflowClient

# 使用MLflow获取刚完成的任务
run_id = 'musing_steelpan_xxxx'

finished_mlflow_run = MlflowClient().get_run(run_id)

报错信息 #

MlflowException                           Traceback (most recent call last)
Cell In [5], line 6
      3 # 使用MLflow获取刚完成的任务
      4 run_id = 'musing_steelpan_hnlbhxf9qy'
----> 6 finished_mlflow_run = MlflowClient().get_run(run_id)

File /miniconda/envs/benchmark/lib/python3.8/site-packages/mlflow/tracking/client.py:150, in MlflowClient.get_run(self, run_id)
    112 def get_run(self, run_id: str) -> Run:
    113     """
    114     Fetch the run from backend store. The resulting :py:class:`Run <mlflow.entities.Run>`
    115     contains a collection of run metadata -- :py:class:`RunInfo <mlflow.entities.RunInfo>`,
   (...)
    148         status: FINISHED
    149     """
--> 150     return self._tracking_client.get_run(run_id)

File /miniconda/envs/benchmark/lib/python3.8/site-packages/mlflow/tracking/_tracking_service/client.py:72, in TrackingServiceClient.get_run(self, run_id)
     58 """
     59 Fetch the run from backend store. The resulting :py:class:`Run <mlflow.entities.Run>`
     60 contains a collection of run metadata -- :py:class:`RunInfo <mlflow.entities.RunInfo>`,
  (...)
     69          raises an exception.
     70 """
     71 _validate_run_id(run_id)
   ...
    648     )
    649 run_info = self._get_run_info_from_dir(run_dir)
    650 if run_info.experiment_id != exp_id:

MlflowException: Run 'musing_steelpan_xxxx' not found

1. 关联MLflow客户端到Azure ML工作区 #

默认初始化的MlflowClient可能指向本地或非目标MLflow服务器，而非你的Azure ML工作区。需显式设置Azure ML工作区的MLflow跟踪URI：

from azure.ai.ml import MLClient
from azure.identity import DefaultAzureCredential
import mlflow
from mlflow.tracking import MlflowClient

# 初始化Azure ML工作区客户端
ml_client = MLClient(
    DefaultAzureCredential(),
    subscription_id="你的订阅ID",
    resource_group_name="你的资源组名称",
    workspace_name="你的工作区名称"
)

# 获取并设置工作区的MLflow跟踪URI
tracking_uri = ml_client.workspaces.get(ml_client.workspace_name).mlflow_tracking_uri
mlflow.set_tracking_uri(tracking_uri)

# 现在获取目标Run
run_id = 'musing_steelpan_xxxx'
finished_mlflow_run = MlflowClient().get_run(run_id)

2. 确认实验上下文匹配 #

Azure ML中的Run隶属于特定实验，如果MLflow客户端默认的实验与目标Run所属实验不一致，也会出现找不到的情况。可以显式指定实验ID或名称：

# 方法1：设置默认实验
mlflow.set_experiment(experiment_name="你的实验名称")

# 方法2：验证Run所属实验ID
client = MlflowClient()
run = client.get_run(run_id)
print(f"Run所属实验ID: {run.info.experiment_id}")

3. 清理MLflow本地缓存 #

本地MLflow配置缓存可能指向旧的跟踪服务器，可按以下操作重置：

• 删除本地mlruns目录（如果存在）
• 重启Python环境后重新执行代码

4. 验证Run的状态与归属 #

在Azure ML门户中确认：

• 目标Run确实存在于当前工作区
• Run状态为已完成
• Run的ID与代码中使用的完全一致（注意大小写和特殊字符）

内容的提问来源于stack exchange，提问作者Simón Cerda