AzureCosmosDB / document-vector-pipeline

Pipeline for ingesting documents (like pdfs and docx) into a searchable Azure Cosmos DB container for vector and hybrid searching.
MIT License
4 stars 3 forks source link

Suggestions #5

Open markjbrown opened 2 weeks ago

markjbrown commented 2 weeks ago

You can simplify the deployment experience for the user and save a couple of steps the user needs to complete manually.

  1. Deploy the code into the azure function directly from bicep instead of manually deploying in step 5.
  2. Tracing can also be setup in the bicep eliminating step 6.
Pilchie commented 2 weeks ago

@amisi01 tried to deploy the azure function from bicep, but couldn't get it to work. Any pointers to how to get this to work?

markjbrown commented 2 weeks ago

You have to use azd to make that work. This github repo shows how to do it. https://github.com/Azure-Samples/functions-quickstart-dotnet-azd and includes nearly everything you need. You will need to add the non-function specific stuff but includes functions, identity, storage. Plus stuff you probably dont need: vnet, private endpoints, etc.

Also, make sure to copy azure.yaml into the root of the project.