Forbidden error with vector embedding skill in Azure AI Search Index

Question

Hi,

I'm currently working with the Azure AI search index and trying to implement vector embedding skills as outlined in the GitHub repository at https://github.com/Azure/azure-search-vector-samples.git. Despite following all the provided steps, I'm encountering a 403 Forbidden error specifically when I try to use the vectorization endpoint. The full error message is as follows: "Could not complete vectorization action. The vectorization endpoint returned status code '403' (Forbidden)." Here are some additional details:

AI search and OpenAI are in private endpoint.

I can successfully create a semantic index, which suggests that the issue is specific to the vectorization process. I'm unsure what might be causing this error or how to troubleshoot it effectively. Has anyone else experienced a similar issue, or does anyone have insights into potential causes or solutions? Any help or guidance on this matter would be greatly appreciated. Thank you!

Answer 1

Hi @A Aathithya this was a known issue and based on this Stack Overflow threadthe issue will be resolved in a week. This was an issue in the current version of the Search and AOAI service communication.

Here are two workarounds that can help:

Turn off trusted access in firewall rules

For a S2 service, use SPL (documentation: Connect through a shared private link - Azure AI Search) to connect to AOAI.

For a service that is not S2, add the search IP address to the AOAI firewall rules (documentation: Connect through firewalls - Azure AI Search)

Best,

Grace