使用自定义泊坞窗映像时，管道运行失败 [英] Pipeline run failed when using custom docker image

问题描述

我在Azure ML管道中使用了自定义泊坞窗图像，并出现以下错误：

回溯（最近一次调用最后一次）：
文件" / mnt / batch / tasks / shared / LS_root / jobs / yipeng-test / azureml / ec4d6d97238b4637a50c981ec377ea42_a3ad8815_2-1-2019_10-45-23_pm / mounts / azureml_project_share / azureml /container_control_script_ec4d6d97238b4637a50c981ec377ea42_a3ad8815_2-1-2019_10-45-23_PM.py" ;,第259行，< module> 
 telemetry_client = get_telemetry_client（）
文件" / mnt / batch / tasks / shared / LS_root / jobs / yipeng-test / azureml / ec4d6d97238b4637a50c981ec377ea42_a3ad8815_2-1-2019_10-45-23_pm / mounts / azureml_project_share / azureml /ec4d6d97238b4637a50c981ec377ea42_a3ad8815_2-1-2019_10-45-23_PM/azureml-setup/utility_context_managers.py" ;,第25行，在get_telemetry_client 
 client = TimeoutBasedTelemetryClient（key）
文件" / mnt / batch / tasks / shared /LS_root/jobs/yipeng-test/azureml/ec4d6d97238b4637a50c981ec377ea42_a3ad8815_2-1-2019_10-45-23_pm/mounts/azureml_project_share/azureml/ec4d6d97238b4637a50c981ec377ea42_a3ad8815_2-1-2019_10-45-23_PM/azureml-setup/utility_context_managers.py" ;,线82，在__init__ 
 self._initialize_appinsight_telemetry_client（appinsight_key）
文件" / mnt / batch / tasks / shared / LS_root / jobs / yipeng-test / azureml / ec4d6d97238b4637a50c981ec377ea42_a3ad8815_2-1-2019_10-45-23_pm / mounts / azureml_project_share / azureml / ec4d6d97238b4637a50c981ec 377ea42_a3ad8815_2-1-2019_10-45-23_PM / azureml-setup / utility_context_managers.py"，第53行，在包装器中
 name =" AppInsightsTelemetryCall"，daemon = True）
 TypeError：__ init __（）得到了意外的关键字参数'守护进程'

使用默认的docker镜像时，例如DEFAULT_CPU_IMAGE（run_config.environment.docker.base_image =  DEFAULT_CPU_IMAGE），一切正常。有趣的是，在这种情况下，AppInsightsTelemetry由于某种原因被禁用：

 
警告：无法实例化AppInsights遥测客户端。遥测已禁用。
 
以下是我正在使用的设置：
来自azureml.core.runco​​nfig import（RunConfiguration，DEFAULT_CPU_IMAGE）
 
 run_config = RunConfiguration（framework =" python"）
 run_config.target = aml_compute 
 run_config.history.output_collection = False 
 run_config.environment.docker.enabled = True 
 run_config.environment.docker.base_image =" google / cloud-sdk" 
 run_config.environment.python.user_managed_dependencies = True 
 
 
我能在这做什么吗？
 
 
 

解决方案

基于错误：

TypeError：__ init__ （）得到一个意外的关键字参数'守护进程'

看起来python2用于docker图像'google / cloud -sdk"。请注意，当前的AML运行时需要python 3环境。

谢谢

I used a custom docker image in my Azure ML pipeline, and got the following error:

Traceback (most recent call last):
  File "/mnt/batch/tasks/shared/LS_root/jobs/yipeng-test/azureml/ec4d6d97238b4637a50c981ec377ea42_a3ad8815_2-1-2019_10-45-23_pm/mounts/azureml_project_share/azureml/container_control_script_ec4d6d97238b4637a50c981ec377ea42_a3ad8815_2-1-2019_10-45-23_PM.py", line 259, in <module>
    telemetry_client = get_telemetry_client()
  File "/mnt/batch/tasks/shared/LS_root/jobs/yipeng-test/azureml/ec4d6d97238b4637a50c981ec377ea42_a3ad8815_2-1-2019_10-45-23_pm/mounts/azureml_project_share/azureml/ec4d6d97238b4637a50c981ec377ea42_a3ad8815_2-1-2019_10-45-23_PM/azureml-setup/utility_context_managers.py", line 25, in get_telemetry_client
    client = TimeoutBasedTelemetryClient(key)
  File "/mnt/batch/tasks/shared/LS_root/jobs/yipeng-test/azureml/ec4d6d97238b4637a50c981ec377ea42_a3ad8815_2-1-2019_10-45-23_pm/mounts/azureml_project_share/azureml/ec4d6d97238b4637a50c981ec377ea42_a3ad8815_2-1-2019_10-45-23_PM/azureml-setup/utility_context_managers.py", line 82, in __init__
    self._initialize_appinsight_telemetry_client(appinsight_key)
  File "/mnt/batch/tasks/shared/LS_root/jobs/yipeng-test/azureml/ec4d6d97238b4637a50c981ec377ea42_a3ad8815_2-1-2019_10-45-23_pm/mounts/azureml_project_share/azureml/ec4d6d97238b4637a50c981ec377ea42_a3ad8815_2-1-2019_10-45-23_PM/azureml-setup/utility_context_managers.py", line 53, in wrapper
    name="AppInsightsTelemetryCall", daemon=True)
TypeError: __init__() got an unexpected keyword argument 'daemon'

When using default docker image, such as DEFAULT_CPU_IMAGE (run_config.environment.docker.base_image = DEFAULT_CPU_IMAGE), things work fine. Interestingly, in that case, the AppInsightsTelemetry is disabled for some reason:

Warning: Couldn't instantiate AppInsights telemetry client. Telemetry disabled.
Here is the settings I am using:
from azureml.core.runconfig import (RunConfiguration, DEFAULT_CPU_IMAGE)

run_config = RunConfiguration(framework="python")
run_config.target = aml_compute
run_config.history.output_collection = False
run_config.environment.docker.enabled = True
run_config.environment.docker.base_image = "google/cloud-sdk"
run_config.environment.python.user_managed_dependencies = True
Anything I can do here?

解决方案

Hi,

Based on the error:

TypeError: __init__() got an unexpected keyword argument 'daemon'

It looks like python2 is used in the docker image 'google/cloud-sdk'. Please note that current AML runtime requires python 3 environment.

Thanks

这篇关于使用自定义泊坞窗映像时，管道运行失败的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！