Skip to content

Conversation

@linusseelinger
Copy link
Contributor

Description of changes

We provide an MLflow server in P4D now, passing the TESSERACT_MLFLOW_TRACKING_URI env var to every Tesseract. With https://github.com/pasteurlabs/AutoPhysics/pull/127, AP follows suit.

This leads to an issue atm if Tesseracts don't explicitly include the mlflow package as a dependency; they receive the env var, switch to mlflow backend, and fail to find the package.

This PR includes mlflow by default in Tesseracts (including cleanups around that).

An alternative could be graceful failure in the scenario above. In that case, it may not be immediately clear to a user why mpa logs of some Tesseracts don't end up in P4D's/AP's MLflow instance. Since have now decided on MLflow as the technology of choice for metrics/params in our ecosystem, we might as well make Tesseracts "batteries included" in that regard.

Testing done

Local testing, CI

@codecov
Copy link

codecov bot commented Dec 18, 2025

Codecov Report

❌ Patch coverage is 16.66667% with 5 lines in your changes missing coverage. Please review.
✅ Project coverage is 39.07%. Comparing base (5284bb6) to head (ceead88).

Files with missing lines Patch % Lines
tesseract_core/runtime/mpa.py 16.66% 5 Missing ⚠️
Additional details and impacted files
@@             Coverage Diff             @@
##             main     #428       +/-   ##
===========================================
- Coverage   66.91%   39.07%   -27.84%     
===========================================
  Files          29       25        -4     
  Lines        3431     3324      -107     
  Branches      535      533        -2     
===========================================
- Hits         2296     1299      -997     
- Misses        945     1917      +972     
+ Partials      190      108       -82     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:
  • ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

@linusseelinger linusseelinger changed the title feat:add mlflow as default feat: add mlflow as default Dec 18, 2025
@linusseelinger linusseelinger force-pushed the linus/feat-mlflow-default-dep branch from ceead88 to 4b7b9bf Compare December 19, 2025 14:47
Author: Dion Häfner <dion.haefner@simulation.science>
@linusseelinger
Copy link
Contributor Author

Cherry-picked changes from #430 , especially for mflow server spun up during CI runs

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants