fix: pin mlflow<2.19 in cli-jobs-pipelines-with-components-pipeline_with_hyperparameter_sweep-pipeline - #3889
Merged
Conversation
… predict_step failure The sklearn-1.5/labels/latest curated environment updated to MLflow 2.19+ around Dec 16 2025. MLflow 2.19 uses the logged-models API which is not supported by AzureML tracking server, breaking mlflow.autolog() in the train step and mlflow.sklearn.load_model() in the predict step. Pin mlflow<2.19 at runtime in both train.yml and predict.yml commands.
Chakradhar886
requested review from
a team,
achauhan-scc,
imatiach-msft,
jayesh-tanna and
kingernupur
as code owners
April 20, 2026 05:45
achauhan-scc
approved these changes
Apr 22, 2026
kingernupur
previously requested changes
Apr 22, 2026
kingernupur
left a comment
kingernupur
left a comment
Member
There was a problem hiding this comment.
@Chakradhar886 Thanks for the detailed explanation. It seems like this notebook is surfacing a deeper problem with our tracking server being incompatible with newer mlflow version.
@jayesh-tanna Do we need to work with the MLFlow service team to fix this instead of constraining the mlflow version here?
lavakumarrepala
changed the title
jayesh-tanna
approved these changes
May 7, 2026
lavakumarrepala
dismissed
kingernupur’s stale review
May 7, 2026 05:36
Jayesh confirmed it is not required to work with mlflow
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
6 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
Problem
The pipeline_with_hyperparameter_sweep notebook has been failing since December 16, 2025 with a JobException on the /score_data pipeline step:
Root Cause
Around December 16, 2025, the curated environment sklearn-1.5/labels/latest was updated to include MLflow 2.19+. This version introduced the logged-models API, which is not supported by the AzureML tracking server.
This breaks two things in the pipeline:
Train step: mlflow.autolog() fails to properly log/save the model via the AzureML backend
Predict/score_data step: mlflow.sklearn.load_model() fails to load the model output from the sweep step
The train step appeared to succeed but produced corrupted or incompatible model artifacts, causing the downstream score_data step to fail.
Fix
Pinned mlflow<2.19 at runtime in both component YAML commands:
train.yml — Added pip install 'mlflow<2.19' -q && before python train.py
predict.yml — Added pip install 'mlflow<2.19' -q && before python predict.py
This downgrades MLflow to a compatible version before executing the scripts, restoring compatibility with the AzureML tracking server.
Checklist