Running Dask dashboard on ARM

I spent last week researching viable alternatives to Spark and came across Dask framework that is designed to run complex data science tasks in parallel.
To demonstrate the power of Dask I used two old workstations running Ubuntu 16.04 to build a small Dask cluster. Each workstation acts as a Dask node with primary objective to run one or more dask-worker processes.
To run the dask-scheduler I used an old Raspberry Pi ARM device. The role of the dask-scheduler is to schedule and distribute tasks to workers, hence low powered Raspberry Pi might have adequate power to run the scheduler.

The only thing that I struggeled with is starting the Dashboard on Raspberry Pi. Dask scheduler outputs no errors and looking thought Dask source code did not help.

I found some hints that the issue might be caused by Tornado and/or Bokeh, but there were no solutions posted.

Well, it turned out that my Raspberry Pi did not have one of the reqired libraries - Library of Linear Algebra routines . Apparently bokeh needs this library.

To get this libray installed:
sudo apt-get install libatlas-base-dev

After installing the library, run the dask-worker and you should see the Dask dashboard.