Error in HNSW Vector Search When Using ORDER BY with Non-Vector Fields in StarRocks

[sharfy]

Steps to reproduce the behavior (Required)
Create the table
CREATE TABLE t_test_vector_table3 (
id BIGINT,
content_embedding ARRAY NOT NULL COMMENT "",
create_time DATETIME,
INDEX func_embedding1 (content_embedding) USING VECTOR (
"dim" = "5",
"index_type" = "hnsw",
"metric_type" = "l2_distance",
"efSearch" = "40",
"M" = "32",
"is_vector_normed" = "false"
)
)
PRIMARY KEY (id)
PROPERTIES (
"replication_num" = "1"
);
Insert sample data
INSERT INTO t_test_vector_table3 (id, content_embedding, create_time)
VALUES (1, [4, 5, 6, 7, 8], NOW());

INSERT INTO t_test_vector_table3 (id, content_embedding, create_time)
VALUES (2, [4, 5, 6, 7, 20], NOW());
Run the query
SELECT id, content_embedding, create_time
FROM t_test_vector_table3
ORDER BY create_time, approx_l2_distance([1, 1, 1, 1, 1], content_embedding)
LIMIT 1;
This query results in the following error:
Expected behavior (Required)
The query should return the nearest vector to [1, 1, 1, 1, 1] while also sorting by create_time.
No errors should occur when including create_time in the ORDER BY clause.
Real behavior (Required)
The query fails with the following error:
Internal error:[2024-12-07 16:17:34] /root/tenann/tenann/searcher/faiss_hnsw_ann_searcher.cc:304:
Error: Error in virtual void faiss::IndexHNSW::search(faiss::Index::idx_t, const float*,
faiss::Index::idx_t, float*, faiss::Index::idx_t*, const faiss::SearchParameters*) const at
/var/local/thirdparty/src/faiss-1.7.3/faiss/IndexHNSW.cpp:291: Error: 'k > 0' failed

StarRocks version (Required)

StarRocks 3.4