We are trying to implement Pega Knowledge buddy in Labs environment. We have created the buddy, data source(using regular Pega Knowledge) and ingested the same.
For all the request(question API) we are getting the response as “I don’t know” from Knowledge buddy.
On further analysis we got the below error in the logs.
Can someone help us understand the issue here.
Error:
com.pega.platform.genai.vectorindex.VectorIndexException: Received status 500 from ‘queryChunks’ for query [User Consent Management] and search options: SearchOptions{k=5, minDistance=0.7, limitToTopChunkPerObject=false, attributesToRetrieve=[roles, dataSource], filterAttributes=Attributes{attributes={roles=Attribute{name=roles, values=[KnowledgeBuddy:Admin, KnowledgeBuddy:Public], type=string, operator=in}, dataSource=Attribute{name=dataSource, values=[knowledge_KC-2086, Knowledge_KC-2086, Knowledge_KC-2085, Knowledge_KCP-5161], type=string, operator=in}}}}. Error: {“code”:500,“message”:“error while querying chunks: ERROR: relation "i_pega_knowledge_embeddings" does not exist (SQLSTATE 42P01)”}
If you are licensed then you will be provisioned with a vector store service which is a requirement for using Buddy.
This is a GenAI-powered tool. All generated answers require validation against the provided references.
It seems like you are encountering a ‘VectorIndexException’ with a status 500 error while using the Question API. This error indicates that there is a configuration problem related to the vector index. Specifically the error message ‘relation i pega knowledge embeddings does not exist’ suggests that the necessary embeddings table is missing or not properly configured in your database. Here are a few steps to troubleshoot and resolve this issue:
Check Data Source Manager Role: Ensure that the user has been assigned the Data Source Manager role. This role is necessary for managing data sources and might be causing the issue if not assigned.
Verify Knowledge Buddy Roles: Make sure that the user has the appropriate Knowledge Buddy roles assigned. Without the necessary roles the user might not have access to the ingested content.
Contact Cloud Team: If the issue persists it might be a deeper configuration problem. In such cases it is recommended to contact the Cloud Team for further assistance.
Trace the Code: If all else fails trace the code to identify any potential issues that might be causing the error.
This is a GenAI-powered tool. All generated answers require validation against the provided references.