Understanding Dask Dashboard

Hello fellow Dask Users!

I’m just starting to mess around with Dask on a cool computer in our campus cluster and I’ve tried using the Dask Dashboard for actually benchmarking things while I run it. Unfortunately, I have little to no idea about what it is the dashboard is telling me!

Is there a handy guide that can walk me through what all the different metrics are for the dashboard and how to tell if what I’m doing is performing properly? I’m pretty new to dask and parallel computing but I’m hoping to use it for speeding up the processing of all the larger datasets my lab is currently collecting (and also because it’s really cool!).

Any advice?

Hi @jmdelahanty welcome to the Dask World. There is documentation about the dashboard and some videos that will explain to you what you see in the dashboard. You can find that information on this documentation page: Diagnostics (distributed) — Dask documentation

This documentation covers the main page of the dashboard, it is a bit out of date but they have there is a great coverage on the dashboard.

To understand how the newer memory plots work, you can check this blogpost that has nice explanations. Tackling unmanaged memory with Dask | Coiled : Coiled

I hope this clarifies some things out if you need information about a specific plot, and or wonder if there is a plot about something you can check the More tab on the dashboard, and if things are not clear feel free to ask.


Thanks to @ncclementi, Dask now has better documentation about the dashboard here: Dashboard Diagnostics — Dask documentation! :sparkles:

1 Like