Fix quickstart MLflow deploy: add azureml-ai-monitoring to model cond…#3923
Open
Chakradhar886 wants to merge 25 commits into
Open
Fix quickstart MLflow deploy: add azureml-ai-monitoring to model cond…#3923Chakradhar886 wants to merge 25 commits into
Chakradhar886 wants to merge 25 commits into
Conversation
…a_env The MLflow no-code online deployment uses an auto-generated score script that imports azureml.ai.monitoring.Collector. Because the notebook overrides MLflow's auto-detected conda_env with a fully custom one, that package was missing in the inference image, causing the worker to crash with ModuleNotFoundError: No module named 'azureml' and liveness/readiness probes returning HTTP 502. Also relax 'pip<=21.3.1' to 'pip' so modern pip is used to resolve the dependencies.
…n't support CreateLoggedModel API)
lavakumarrepala
added a commit
that referenced
this pull request
May 7, 2026
- Unpin pip (pip<=21.3.1 -> pip) to allow newer package installs - Add azureml-ai-monitoring to model conda env for deployment liveness probe - Pin mlflow<3 and mlflow-skinny<3 at runtime to avoid logged-models API 404 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…oy-model" This reverts commit 6817708.
…(sklearn 1.5.2, mlflow 2.17.0, numpy<2)
Chakradhar886
force-pushed
the
fix/quickstart-mlflow-deploy-azureml-ai-monitoring
branch
from
66aa4e7 to
78a0ec1
Compare
…azureml-ai-monitoring, psutil, pin pandas==1.5.3)
…rt (add azureml-ai-monitoring, psutil, pin pandas==1.5.3)" This reverts commit 7126f33.
…ntainer (sklearn 1.5.2, mlflow 2.17.0, numpy<2)" This reverts commit 78a0ec1.
…de inference image Workaround for inference-base-2204 image whose auto-injected mlflow_score_script.py imports azureml.ai.monitoring (not installed in baked env), causing all no-code MLflow online deployments to crashloop. Switch deploy-model.ipynb's blue/green deployments to a custom scoring script and a custom Environment built from inference-env.yaml so the broken auto-injected script is never used. Also re-applies the regenerated credit_defaults_model artifact (sklearn 1.5.2 / mlflow 2.17.0) so mlflow.pyfunc.load_model can deserialize it under the modern stack.
…ow no-code inference image" This reverts commit afd9ef4.
…low no-code inference image" This reverts commit 11b825c.
…oken MLflow no-code inference image"" This reverts commit 1c85776.
…da.yaml (minimal fix for inference image 502)
… inference server ModuleNotFoundError: azureml.contrib)
…ainer 502: ModuleNotFoundError azureml.contrib)
…ces in training cmd so MLflow infers them into model requirements (fix container 502: ModuleNotFoundError azureml)
…itoring & azureml-contrib-services are written into model requirements.txt (MLflow no-code deploy uses requirements.txt, not conda_env)
…v); rely on conda_env which already lists azureml-ai-monitoring & azureml-contrib-services
…() lookup, so deployment always uses the static model just registered (avoids picking up a quickstart-registered version that lacks azureml packages)
…match new inference base image (py3.8 unsupported); resolves liveness probe 502 when deploying the static credit_defaults_model
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
1 participant
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.
…a_env
The MLflow no-code online deployment uses an auto-generated score script that imports azureml.ai.monitoring.Collector. Because the notebook overrides MLflow's auto-detected conda_env with a fully custom one, that package was missing in the inference image, causing the worker to crash with ModuleNotFoundError: No module named 'azureml' and liveness/readiness probes returning HTTP 502.
Also relax 'pip<=21.3.1' to 'pip' so modern pip is used to resolve the dependencies.
Description
Checklist