Unable to connect to self-hosted weaviate clusters through verba 2.0 version

Support
1

I am trying to access verba 2.0 version with weaviate cluster 1.26.3 version.

export WEAVIATE_API_KEY_VERBA=“pwd”
export OLLAMA_EMBED_MODEL=llama3.1
export OLLAMA_MODEL=llama3.1
export WEAVIATE_URL_VERBA=https://:6383
export OLLAMA_URL=http://ollama-url:2300

Error :
https://::6383 with Auth
✘ Couldn’t connect to Weaviate, check your URL/API KEY: Invalid port:
‘6383:443’
✘ Failed to connect to Weaviate Couldn’t connect to Weaviate, check
your URL/API KEY: Invalid port: ‘6383:443’
INFO: 10.82.204.17:65400 - “POST /api/connect HTTP/1.1” 400 Bad Request

image
image3380×1454 214 KB

note : It was working fine with verba 1.0 version !!

Please let me know how to fix this issue ?

2

Hi @adithya.ch !

If you are using it locally, hosting using the embeddedm you need to choose local.

This option will work for Weaviate hosted in our cloud.

3

Hi @DudaNogueira ,

We hosted weaviate cluster locally on our kubernetes cluster.

It was working fine with v1.0 ? Was there any process/breaking changes with V2.0 disabling access to self hosted weaviate cluster ?

Regards,
Adithya

4

Right now the new Verba 2.0 will give you three options to connect:

  • Local (Embedded, will spin a Weaviate instance)
  • Weaviate (will allow you to provide URL and API KEY to connect to our cloud)
  • Docker. will use connect_to_local. It expects a host named weaviate at http port 8080 and grpc port 50051.

As you are using K8s, you will need provide Verba pod a way to connect to your K8s Weaviate instance having those exact informations.