Save file locally in jupyterhub notebook running on EMR cluster

Question

When using Jupyterhub application interface (via SSH tunneling) on Amazon EMR, the default file explorer says /user/jovyan/tree. What directory is this and how can I save a file (say a matplotlib figure) from within the notebook to this local space? The usual tricks like

os.chdir(sys.path[0])
print(os.getcwd())

show /mnt/tmp, which does not actually appear to be the local directory.

Answer 1

The code that you are running in EMR notebook is actually saved on s3. By default, it is often in a bucket named something like emr-resources, under the notebooks folder.

Try checking your s3 and I believe if you save a file from within an emr notebook, it will be available save in s3 (and available next time)