Lost Work on Win - jupyter notebook not persisting changes

Hello. Many many times I observed cases of lost work in Jupyter Notebook.
I mostly work on Windows Server 2016 with latest updates/Chrome/Python of latest version, and use latest Anaconda (downloaded & installed a week ago).
Today it happened again, lost several hours of intense work. I was pressing Ctrl+S like every 5 minutes, just a single connection to python was open, not a single indication of problem of any kind was given by Jupyter. As my work was very important, I was additionally backing up my .ipynb file every hour, so I ended up having like 3 backups. Then during computation my notebook hung, it tried stopping calculations, it did not work, i tried restarting kernel - did not work either, icon of dead kernel was shown. Ok does not matter, I thought, killed all python processes and relaunched the Jupyter Notebook. Opened the file, and, surprise, surprise! All changes made during last 3 hours of work are gone. How is that possible at all? Could someone take a look at code responsible for saving? Clearly it's a bug there and not being fixed for at least half year according to my observations.
.ipynb_checkpoints folder contains only 1 file with date of 4 hrs ago.
Cherry on top of the cake: all backups made by me manually each contained the same obsolete version of the file. JpNb was neither updating the file at all not giving any indications of a problem.

If you don't believe me you can easily find people having similar problems:
https://forum.humanbrainproject.eu/t/jupyter-notebooks-saving-fails-silently/143/3

Thank you.

Start by reproducing the reported Windows save failure around Ctrl+S, a hung notebook, kernel restart, and relaunch. Compare the reopened .ipynb file with the file in .ipynb_checkpoints; done means identifying a reliable reproduction and ensuring unsaved changes are persisted or that failure is clearly reported.

Written by the indexing model from the issue text.