Azure / Azure/azure-sdk-for-python
Mlflow tracing not working
Open
Client
customer-reported
feature-request
Machine Learning
needs-team-attention
Service Attention
- Dominant language
- Python
- Stars
- 5.6k
- Forks
- 3.4k
- Avg merge
- 1d 21h
- Merged PRs (30d)
- 193
Description
**Describe the bug**
When trying to use the tracing features of mlflow in AzureML:
https://mlflow.org/docs/latest/tracing/
There is an error on the client side:
mlflow.tracking.client: Failed to start trace AsyncCompletions: API request to endpoint /api/2.0/mlflow/traces failed with error code 403 != 200. Response body
**To Reproduce**
Steps to reproduce the behavior:
1. You just need to setup mlflow as usual with the right tracking uri.
2. Add the line: mlflow.openai.autolog()
**Expected behavior**
It shouldn't crash the code as autolog is quite standard, even if tracing isn't supported, it should at the very least save the traces somewhere.
Contributor guide
Assessment
This issue has not been assessed yet.