Forum Discussion
Lost Notebook Changes in Synapse Data Science (Microsoft Fabric)
Chiming in here for lack of better ideas. I've also had some issues with losing notebook work in Fabric. Possibly for different reasons (I think mine were induced by having multiple connections to the same notebook accidentally in a hybrid work accident), but I find the work recovery abilities for notebooks in Fabric subpar. I submitted an 'idea' for Fabric to automatically and/or on demand checkpoint notebooks. This is already a default feature of Jupyter Notebooks and is integrated with notebooks in Azure Machine Learning, but is missing in Fabric notebooks.
AML has a description of its checkpoint behaviour here: Run Jupyter notebooks in your workspace - Azure Machine Learning | Microsoft Learn . I don't see why this wouldn't be implemented for Fabric as well.
I've faced this issue once myself, and I think I know the root cause. Notebooks will autosave in Fabric, and if you have multiple windows or tabs open in your browser that both have the notebook open, even if you update the notebook in Tab A, if you flip back to Tab B in your browser that has the notebook open, it will autosave that old version of your notebook over the version you've been developing in Tab A, as they both point to the same Notebook in the back-end.
That was my experience as when I stopped using mutiple broswer tabs I never re-encountered the issue, so I'm working under that assumption.
