Use Jupyter Notebooks to Help People on StackOverFlow

I am sat in the PowerShell Saturday in Hamburg. You can see me on the right of this picture writing my previous blog post!

I was talking with my friend Mathias Jessen @IISResetMe on Twitter about notebooks and he said that another great use case was to use them on Stack OverFlow

Now Mathias is an active answerer on Stack OverFlow

and he puts a lot of effort into writing his answers, formatting them, including code and results. Basically exactly the same as a Notebook. However, with a Notebook, you can enable people to run the code as well on their own machines.

Mathias says he will use notebooks to help people when he answers their PowerShell questions on Stack OverFlow. If you are a Stack OverFlow Answerer then you can too.

.NET PowerShell Notebooks – Using Pester

My last post had a lot of information about the new .NET PowerShell notebooks including installation instructions.

.NET Notebooks are Jupyter Notebooks that use .NET core to enable C#, F# and PowerShell kernels.

Use Cases

One of the main benefits that I see for Jupyter Notebooks for Ops folk is that the results of the query are saved with the notebook. This makes them fantastic for Incident resolution.

If you have an incident at 3am and you know that you will need that information in the wash up meeting the next day instead of copying and pasting results into a OneNote document or a text file, you can simply run the queries in a notebook and save it.

In the meeting, you can simply open the notebook and the results will be available for everyone to see.

Even better, if you have a template notebook for those scenarios and you can then compare them to previous occurrences.

Using Pester

Using Pester to validate that an environment is as you expect it is a good resource for incident resolution, potentially enabling you to quickly establish an area to concentrate on for the issue. However, if you try to run Pester in a .NET Notebook you will receive an error

Describe: 
Line |
   3 | Describe "Checking Problem ...... by $($ENV:USERDOMAIN) $($ENV:UserName)" {

     | ^ The 'Describe' command was found in the module 'Pester', but the module could not be loaded. For more information, run 'Import-Module Pester'.

Fixing it

When you try to Import-Module Pester you get the following error

Get-Command: C:\Users\mrrob\Documents\PowerShell\Modules\Pester\4.9.0\Pester.psm1
Line |
  94 |     $script:SafeCommands['Get-CimInstance'] = Get-Command -Name Get-CimInstance -Module CimCmdlets @safeCommandLookupParameters

     |                                               ^ The term 'Get-CimInstance' is not recognized as the name of a
     | cmdlet, function, script file, or operable program. Check the spelling of the name, or if a path was included,
     | verify that the path is correct and try again.
 Import-Module: The module to process 'Pester.psm1', listed in field 'ModuleToProcess/RootModule' of module manifest 'C:\Users\mrrob\Documents\PowerShell\Modules\Pester\4.9.0\Pester.psd1' was not processed because no valid module was found in any module directory. 

Thats odd, why is it failing there? Dongbo Wang from the PowerShell team explains in the issue that I raised

Yes, it was the CimCmdlets module from the system32 module path that got imported (via the WinCompat feature added in PS7). This is because currently the PS kernel don’t ship all the built-in modules along with it …
The built-in modules are not published anywhere and are platform specific, it’s hard for an application that host powershell to ship them along. We have the issue PowerShell/PowerShell#11783 to track this work.

The way to resolve this is to Import the CimCmdlets Module from your local PowerShell 7 installation until the issue is resolved

Import-Module 'C:\program files\powershell\7-preview\Modules\CimCmdlets\CimCmdlets.psd1'

Then you can run your Pester

You can see all of this including all the results in this notebook that I have created and shared on Github and also below as a gist to embed in this blogpost

Sharing Code AND Results 🙂


Notebooks – A brilliant way of sharing what you did and the results that you got enabling others to follow along. You can do this with this Notebook. Download it and open it in your Jupyter Lab and you will be able to run it and see all of the errors and the fix on your machine.

New .NET Notebooks are here – PowerShell 7 notebooks are here.

Data Science folk used Notebooks for documentation and to show re-runnable research. Azure Data Studio included this notebook functionality and added SQL kernel where with a little bit of faffing you could run PowerShell and then a Python kernel that enabled PowerShell. It seems that notebooks are so cool that everyone is creating them these days! I was browsing twitter when I saw this tweet.

PowerShell 7 Notebooks 🙂

A notebook experience for PowerShell 7 that sounds amazing. This will enable a true cross-platform PowerShell Notebook experience which is lacking from the Python version as it uses Windows PowerShell on Windows and PowerShell Core on other OS’s

The first thing I asked was – Will this come to Azure Data Studio. I got an immediate response from Sydney Smith PowerShell Program Manager saying it is on the roadmap

Install dependencies

To be able to run the notebook, you need to install some dependencies. First install the .NET CORE SDK which you can download from https://dotnet.microsoft.com/download This needs admin permissions to install.

You also need a Python installation – You can use Anaconda, which you can download from here https://www.anaconda.com/distribution/ This does not need admin to install

Add Anaconda to Windows Terminal

I have added the Anaconda prompt to Windows Terminal so that I have one entry point for all my CLIs. Open the settings file and add the code below. (It will also give you an icon and background.

        {
            // Make changes here to the Anaconda.exe profile
            "guid": "{0caa0dad-35be-5f56-a7ff-afceeeaa6101}",
            "name": "Anaconda",
            "commandline": "cmd.exe /K C:\\Users\\mrrob\\Anaconda3\\Scripts\\activate.bat",
            "hidden": false,
            "backgroundImage": "C:\\Users\\mrrob\\Anaconda3\\Menu\\anaconda-navigator.ico",
            "icon": "C:\\Users\\mrrob\\Anaconda3\\Menu\\anaconda-navigator.ico",
            "backgroundImageAlignment": "topRight",
            "backgroundImageStretchMode": "uniform",
            "backgroundImageOpacity": 0.1
        }

and it appears in the drop down

With Anaconda installed, check that that the kernel is available on your path. If like me you have Azure Data Studio installed, you will have additional kernels but the important one line here is

python3 C:\Users\USERNAME\Anaconda3\share\jupyter\kernels\python3

In Windows Terminal move to a PowerShell 7 prompt and install the dotnet interactive tool

dotnet tool install --global Microsoft.dotnet-interactive

Then you can install the .NET kernel in your Anaconda prompt using this command

dotnet interactive jupyter install

Sometimes new things have errors

I had an error when I tried this first time

Could not execute because the specified command or file was not found.
Possible reasons for this include:
* You misspelled a built-in dotnet command.
* You intended to execute a .NET Core program, but dotnet-interactive does not exist.
* You intended to run a global tool, but a dotnet-prefixed executable with this name could not be found on the PATH.

This is easily fixed by adding %USERPROFILE%\.dotnet\tools to my path with set PATH=%PATH%;%USERPROFILE%\.dotnet\tools

Running jupyter kernelspec list shows that the .NET kernel is installed for C Sharp, F Sharp and .NET PowerShell

Lets open a Notebook

Now you want to play with it!
You can run the lab environment using `jupyter lab`

This opens a browser

You can open existing Azure Data Studio PowerShell notebooks (but not SQL ones)

Sometimes new things have errors Part 2

Unfortunately, I get errors when trying to import Pester which means I can not use my dbachecks notebooks in this blog post. I have raised an issue on the repo here.

Create a New Notebook

But it is easy to create a new Notebook

In the launcher page click the .NET PowerShell button


Which will open a new Notebook in the directory that you launched the lab from. You can then add Code or Markdown as I have described before here.

Then you can add code, markdown and images to create your notebook.

Once you have finished using the notebook lab, you can shut it down in the Anaconda prompt with CTRL + C

Here is a video of running a notebook which anyone can use to create a couple of Docker containers running SQL 2019 and query them with dbatools. You can find the notebook further down this post.

Sharing Notebooks

You can create notebooks to run common tasks. Even better, from the lab you can convert the notebook including the results to a variety of formats to share with other none-technical people. I used this functionality this week to export Azure Data Studio Notebooks to HTML and PDF for a Project manager 🙂

You can find the Export Notebook command in the File menu

Exporting to HTML did not export the images but it does include the results

You can share notebooks via GitHub – Either in a gist like this

or by providing a straight link to the notebook in GitHub https://github.com/SQLDBAWithABeard/Notebooks/blob/master/notebooks/Exploring%20dbatools.ipynb

You can also use Binder https://mybinder.org/

This uses Docker to create an interactive Notebook. Create a Github repo like https://github.com/SQLDBAWithABeard/Notebooks (or just clone it) Copy your notebooks into the notebooks folder and push the changes to Github and then go to https://mybinder.org/ and add your URL to the repository.

You can see what it looks like by clicking the button below which Binder creates for you

Unfortunately the kernel only supports Python for the moment but you can see the possibilities 🙂