Error when logging to MLFlow deleted experiment #20556
Open
Conversation
This file contains 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
millskyle
requested review from
lantiga,
Borda,
tchaton,
justusschock and
ethanwharris
as code owners
for more information, see https://pre-commit.ci
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
What does this PR do?
Addresses issue #20555.
When logging to an MLFlow experiment name that has been deleted (e.g. deleted from the MLFlow UI), the MLFlow experiment still exists in the backend, so
expt = self._mlflow_client.get_experiment_by_name(self._experiment_name)retrieves it and Lightning proceeds as if the experiment already exists. When Lightning actually tries to log to the experiment, it stalls as MLFlow returns many 500 errors. Eventually it times out and the program crashes.This PR simply modifies the check to ensure that the run exists and also is not deleted. This will cause the experiment creation to fail, but with an error message that explicitly states that it cannot be created because there is a deleted experiment with the same name.
This could be handled differently to not crash the run (e.g. increment the experiment name by post-pending _1, _2, etc.) but I think the explicit failure with a clear message is my preferred behaviour.
Fixes #20555>
📚 Documentation preview 📚: https://pytorch-lightning--20556.org.readthedocs.build/en/20556/