Details
-
Bug
-
Resolution: Unresolved
-
Major
-
None
-
Morpheus
-
None
-
Enterprise Edition 8.0.0 build 1709
-
Untriaged
-
-
0
-
No
Description
creating a index with incorrect value in the with clause leaves a trace of index in the workbench. The dimension should have 3
CREATE INDEX `testmultiScalarOneVector_2` ON b1(rating,category,colorRGBVector Vector)
|
PARTITION BY HASH (meta().id) WITH {"dimension":384, "description": "IVF,PQ8x8", "similarity":"L2", 'num_partition': 8} |
Also the error message has no clear indicator why the index creation failed. The error message has the same repetitive message all over the place as well.
[
|
{
|
"code": 5000, |
"msg": "GSI CreateIndex() - cause: 4:ErrTraining: Number of qualifying/valid vectors 0 are less than the number of vectors 256 required for training 256 centroids 5:ErrTraining: Number of qualifying/valid vectors 0 are less than the number of vectors 256 required for training 256 centroids 6:ErrTraining: Number of qualifying/valid vectors 0 are less than the number of vectors 256 required for training 256 centroids 7:ErrTraining: Number of qualifying/valid vectors 0 are less than the number of vectors 256 required for training 256 centroids 8:ErrTraining: Number of qualifying/valid vectors 0 are less than the number of vectors 256 required for training 256 centroids 1:ErrTraining: Number of qualifying/valid vectors 0 are less than the number of vectors 256 required for training 256 centroids 2:ErrTraining: Number of qualifying/valid vectors 0 are less than the number of vectors 256 required for training 256 centroids 3:ErrTraining: Number of qualifying/valid vectors 0 are less than the number of vectors 256 required for training 256 centroids \n\n", |
"reason": { |
"_level": "exception", |
"caller": "secondary_index:683", |
"cause": { |
"error": "4:ErrTraining: Number of qualifying/valid vectors 0 are less than the number of vectors 256 required for training 256 centroids 5:ErrTraining: Number of qualifying/valid vectors 0 are less than the number of vectors 256 required for training 256 centroids 6:ErrTraining: Number of qualifying/valid vectors 0 are less than the number of vectors 256 required for training 256 centroids 7:ErrTraining: Number of qualifying/valid vectors 0 are less than the number of vectors 256 required for training 256 centroids 8:ErrTraining: Number of qualifying/valid vectors 0 are less than the number of vectors 256 required for training 256 centroids 1:ErrTraining: Number of qualifying/valid vectors 0 are less than the number of vectors 256 required for training 256 centroids 2:ErrTraining: Number of qualifying/valid vectors 0 are less than the number of vectors 256 required for training 256 centroids 3:ErrTraining: Number of qualifying/valid vectors 0 are less than the number of vectors 256 required for training 256 centroids", |
"source": "CreateIndex()" |
},
|
"code": 4350, |
"key": "indexing.error", |
"message": "GSI error" |
},
|
"query": "CREATE INDEX `testmultiScalarOneVector_2` ON b1(rating,category,colorRGBVector Vector)\nPARTITION BY HASH (meta().id) WITH {\"dimension\":384, \"description\": \"IVF,PQ8x8\", \"similarity\":\"L2\", 'num_partition': 8}" |
}
|
]
|
This is the correct index definition and work well
CREATE INDEX `testmultiScalarOneVector_2` ON b1(rating,category,colorRGBVector Vector)
|
PARTITION BY HASH (meta().id) WITH {"dimension":3, "description": "IVF,PQ3x8", "similarity":"L2", 'num_partition': 8} |