Description
I saw an error in production where a vector was not found:
explorer: get class: concurrentTargetVectorSearch): explorer: get class: vectorize search vector: nearObject params: vector not found
Upon further inspection, I noticed the object being queried does not exist when querying via GraphQL:
query {
Get {
ContentOpenAI(where: {
path: ["datagraph_id"]
operator: Equal
valueString: "cqpnth0vub2s5hpce720"
}) {
datagraph_id
datagraph_type
name
}
}
}
However, when querying the object list via REST, it is there:
GET {{baseUrl}}/objects/9f27c258-05d1-5987-bc13-7c39436d3a8a?include=vector
{
"class": "ContentOpenAI",
"creationTimeUnix": 1723041716137,
"id": "9f27c258-05d1-5987-bc13-7c39436d3a8a",
"lastUpdateTimeUnix": 1723041716137,
"properties": {
"content": "Things that make me more money\r\n\r\n",
"datagraph_id": "cqpnth0vub2s5hpce720",
"datagraph_type": "profile",
"description": "Things that make me more money",
"name": "david"
},
"vector": [ ... ]
}
Note how the “datagraph_id” is present in the response, the exact same one used in the GQL query above. I’ve also pulled all objects via GQL and it’s not in the list.
Am I misunderstanding something about Weaviate here? Are there cases where objects appear in one API but not in another? I wondered if it had not indexed correctly via embeddings, but that seems not the case because a full vector is present (omitted from the post for readability.)
Server Setup Information
- Weaviate Server Version: 1.26.1
- Deployment Method: Fly.io
- Client Language and Version: Just hitting GQL/REST directly