向Watson Studio项目添加Machine Learning服务时遇500错误求助

What's Causing This? #

That Error while selecting instance:Unexpected response code: 500 (Missing required properties for credentials: access_key) error happens due to a compatibility gap between the latest IBM Machine Learning service and Watson Studio.

Newer ML service instances no longer include the access_key field in their credentials, but the current version of Watson Studio still expects this field to exist. When it can't locate it, it throws that 500 server error.

Quick Workarounds to Try #

Here are two reliable fixes to get past this issue:

• Manually Add the Missing Field to Credentials

  1. Head to your IBM Cloud Console, find your ML service instance, and go to the Credentials tab. Either create a new credential or copy the JSON of an existing one.
  2. Edit the JSON to add "access_key": "" (you can leave the value empty—Watson Studio just needs the field to exist for validation, the new ML service doesn't actually use it).
  3. Back in Watson Studio, when adding the service, choose Enter credentials manually, paste your modified JSON, and finish the setup.

• Create a Legacy ML Service Instance
  If the manual edit doesn't work, try spinning up a legacy version of the ML service:

  1. When creating a new ML service instance in IBM Cloud, look for the Service version option in the configuration page.
  2. Select the older version (if available) and complete the instance creation.
  3. This legacy instance's credentials will include the access_key field, which Watson Studio can recognize without issues.

Long-Term Fix #

This is a known compatibility bug between IBM's services. The best long-term solution is to report this issue through IBM Cloud's support channels—this will help their team push an update to Watson Studio that supports the new ML service credential structure.

内容的提问来源于stack exchange，提问作者David Carew