Hi FriendlyElk26 ,
I think this is caused by the GCP LB automatically dropping payload for GET requests (no other LB does that). To work around it, you need to make ClearML use PUT requests instead by setting the configuration option api.http.default_method=PUT in the configuration file (or using the CLEARML_API_DEFAULT_REQ_METHOD environment variable. Note that there was a fix for this in the latest version, so I would recommend using the latest versions for both clearml SDK and the agent (for the agent, use the latest RC from pypi)