A wrong query structure of hamming/jaccard space will trigger java.security.AccessControlException and the server will crash

[liuyonghengheng]

Support guidelines

• [x] I've read the support guidelines

Background

step 1:

define a mapping with type elastiknn_sparse_bool_vector and use jaccard space

curl  -H'Content-Type: application/json'  -X  PUT  http://localhost:9200/my-vector-index -d '{
"mappings":{
    "properties": {
        "my_vec1": {
            "type": "elastiknn_sparse_bool_vector",
            "elastiknn": {
                "dims": 10, 
                "model": "lsh",
                "similarity": "jaccard",
                "L": 99,
                "k": 1
            }
        }
}
}}'

step2: using a wrong structure of vec:

curl -H'Content-Type: application/json' -X GET http://localhost:9200/my-vector-index/_search -d'{
    "query": {
        "elastiknn_nearest_neighbors": {        
            "field": "my_vec1", 
            "vec": {
                "values": [1,1,1,1,1,1,1,1,1,1]
            },
            "model": "lsh",
            "similarity": "jaccard",
            "candidates": 50
        }
    }
}'

then we should received a error response，but the node server crashed.

Bug

from the log we can find the reason：the function DenseFloat.toString call the class scala.reflect.ClassTag，but do not have the permission of getClassLoader.

Elastiknn Version

8.17.5.0

Platform

my laptop with system ubuntu22

Steps to reproduce

step 1:

define a mapping with type elastiknn_sparse_bool_vector and use jaccard space

curl  -H'Content-Type: application/json'  -X  PUT  http://localhost:9200/my-vector-index -d '{
"mappings":{
    "properties": {
        "my_vec1": {
            "type": "elastiknn_sparse_bool_vector",
            "elastiknn": {
                "dims": 10, 
                "model": "lsh",
                "similarity": "jaccard",
                "L": 99,
                "k": 1
            }
        }
}
}}'

step2: using a wrong structure of vec:

curl -H'Content-Type: application/json' -X GET http://localhost:9200/my-vector-index/_search -d'{
    "query": {
        "elastiknn_nearest_neighbors": {        
            "field": "my_vec1", 
            "vec": {
                "values": [1,1,1,1,1,1,1,1,1,1]
            },
            "model": "lsh",
            "similarity": "jaccard",
            "candidates": 50
        }
    }
}'

then we should received a error response，but the node server crashed.

Additional info

No response