Vector Search Index Creation Failed With No Errors

Generative AI Vector Search for LLMs
, ,
1

Im getting an error where vector search index creation fails. Im using compass with the following index definition:

{
  "fields": [
    {
      "type": "vector",
      "path": "items.embeddings",
      "numDimensions": 1536,
      "similarity": "cosine"
    }
  ]
}

There is no error log in the database logs. I am running mongodb locally using docker compose.

2

Hi @Omar_Farooq, thank you for posting your issue.

Can you share the full syntax (including the syntax for the atlas cli call itself)? That may help make it more clear what the issue is.

Thanks!

3

Hi @Omar_Farooq ,

Without having more background about your problem, I assume you’re running MongoDB Community edition on Docker. If this is the case, Community Edition does not support Vector Search (does not have Atlas Search). But you can create a local Atlas cluster for development using atlas-cli

Just install atlas-cli and from a command line type:

$ atlas deployments setup

This will guide you through the creation of a local (or Atlas hosted) Atlas Cluster you can use for local development that will support Atlas Vector Search.

4

I have resolved the error. The error was due to incorrect schema. I was trying to create an atlas search index on an array rather than a collection. Thanks anyways.

6

Glad you were able to resolve the issue!