You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
I'm sending a knn query to a number of indices but they don't necessarily all have the knn field
GET my_index_prefix_*/_search
{"query": {"knn": {"embedding_256": {"k": 10, "vector": [......]}}}
fails with this error
{
"shard": 0,
"index": "my_index_prefix_123456",
"node": "xxxxxx",
"reason": {
"type": "query_shard_exception",
"reason": "failed to create query: Field 'embedding_256' is not knn_vector type.",
"index": "my_index_prefix_123456",
"index_uuid": "xxxxxx",
"caused_by": {
"type": "illegal_argument_exception",
"reason": "Field 'embedding_256' is not knn_vector type."
}
}
}
Describe the solution you'd like
Like nested types, knn types can also have an ignore_unmapped field, so I can search among only the indices with this knn field. The query would be like
Is your feature request related to a problem? Please describe.
I'm sending a knn query to a number of indices but they don't necessarily all have the knn field
fails with this error
Describe the solution you'd like
Like nested types, knn types can also have an
ignore_unmapped
field, so I can search among only the indices with this knn field. The query would be likeDescribe alternatives you've considered
Force adding an empty knn field to all indices. But that is troublesome.
The text was updated successfully, but these errors were encountered: