Setup does not set all appropriate values

[jhST]

Context / Scenario

On a fresh check-out of the source, running setup.cmd does not yield a runnable configuration.

What happened?

Config generated by setup Attached.
appsettings.Development.json
When running, output is:

`******
Text generation (TextGeneratorType) is not configured.
Please configure the service and retry.

How to configure the service:

1. Set the ASPNETCORE_ENVIRONMENT env var to "Development" or "Production".

   Current value: Development
   `

Importance

I cannot use Kernel Memory

Platform, Language, Versions

Windows, current main.

Relevant log output

No response

[dluc]

could you share the answers given during the setup?

depending on the answers given, there might be an edge case non properly covered leading to this error.

[NikiBase]

Hi @dluc , I run into the same error. In my case the message displaying after the dotnet run setup is the following:

== Done! :-)

== You can start the service with: dotnet run

After doing dotnet run, this displays:

******
Text generation (TextGeneratorType) is not configured.
Please configure the service and retry.

How to configure the service:

1. Set the ASPNETCORE_ENVIRONMENT env var to "Development" or "Production".

   Current value: Development

2. Manual configuration:

     * Create a configuration file, either "appsettings.Development.json" or
       "appsettings.Production.json", depending on the value of ASPNETCORE_ENVIRONMENT.

     * Copy and customize the default settings from appsettings.json.
       You don't need to copy everything, only the settings you want to change.

  Automatic configuration:

     * You can run `dotnet run setup` to launch a wizard that will guide through
       the creation of a custom "appsettings.Development.json".

  Adding components:

     * If you would like to setup the service to use custom dependencies, such as a
       custom storage or a custom LLM, you should edit Program.cs accordingly, setting
       up your dependencies with the usual .NET dependency injection approach.

******

but the appsettings.Development.json is created during the dotnet run setup
Thank you in advance!

[dluc]

hi @NikiBase I'd need to know the answers given (except for the private data) in order to reproduce the error. Could you share yours? Thanks

[NikiBase]

For each question during the setup phase? I can send you the .json generated with the private data anon, would that be enough? Thanks

[NikiBase]

Hi, @dluc is this enough?

How should Kernel Memory service run and handle memory and documents ingestion?

Web Service with Asynchronous Ingestion Handlers (better for retry logic and long operations)

Protect the web service with API Keys?

No

Enable OpenAPI swagger doc at /swagger/index.html?

No

Which queue service will be used?

SimpleQueues volatile (only for tests, data stored in memory)

Where should the service store files? A persistent storage is required to handle updates, downloads, etc.

SimpleFileStorage volatile (only for tests, data stored in memory)

Which service should be used to extract text from images?

None

When importing data, generate embeddings, or let the memory Db class take care of it?

No, my memory Db class/engine takes care of it

When searching for text and/or answers, which embedding generator should be used for vector search?

None/Custom (manually set with code)

When searching for answers, which memory DB service contains the records to search?

SimpleVectorDb volatile (only for tests, data stored in memory)

When generating answers and synthetic data, which LLM text generator should be used?

None/Custom (manually set with code)

Log level?

Warning

I am using this setup just for test proposes. Thank you!

[dluc]

hi @NikiBase I see now, some of the answers given requires changing the code, ie. dependencies are either defined in the config file or in the code, editing the code. One of the answers is also incorrect.

• When importing data, generate embeddings, or let the memory Db class take care of it?

  • Answer given: No, my memory Db class/engine takes care of it
  • Memory Db selected: SimpleVectorDb volatile (only for tests, data stored in memory)
  • Problem: SimpleVectorDb does not generate embedding automatically.
  • Solutions:
    • Option 1: use Azure AI Search with Semantic Indexing or similar services that can index data semantically without the need to provide an embedding vector
    • Option 2: Answer Yes, and then choose an AI service to generate embeddings

• When searching for text and/or answers, which embedding generator should be used for vector search?

  • Answer given: None/Custom (manually set with code)
  • Problem: the default code doesn't work with this, unless you edit it
  • Solutions:
    • Option 1: edit the code and define an embedding generator in your custom code
    • Option 2: Answer Yes, and then choose an AI service to generate embeddings

• When generating answers and synthetic data, which LLM text generator should be used?

  • Answer given: None/Custom (manually set with code)
  • Problem: the default code doesn't work with this, unless you edit it
  • Solutions:
    • Option 1: edit the code and define a text generator in your custom code
    • Option 2: Answer Yes, and then choose an AI service to generate text