UNSUPPORTED_FEATURE error class
SQLSTATE: 0A000
The feature is not supported:
AES_MODE
AES-<mode>
with the padding <padding>
by the <functionName>
function.
AES_MODE_AAD
<functionName>
with AES-<mode>
does not support additional authenticate data (AAD).
AES_MODE_IV
<functionName>
with AES-<mode>
does not support initialization vectors (IVs).
AI_FUNCTION
AI function <functionName>
is only available in Interactive Workloads, Jobs, SQL Pro and SQL Serverless, or it’s disabled explicitly.
ANALYZE_UNCACHED_TEMP_VIEW
The ANALYZE TABLE FOR COLUMNS command can operate on temporary views that have been cached already. Consider to cache the view <viewName>
.
ANALYZE_UNSUPPORTED_COLUMN_TYPE
The ANALYZE TABLE FOR COLUMNS command does not support the type <columnType>
of the column <columnName>
in the table <tableName>
.
ANALYZE_VIEW
The ANALYZE TABLE command does not support views.
CATALOG_OPERATION
Catalog <catalogName>
does not support <operation>
.
COMBINATION_QUERY_RESULT_CLAUSES
Combination of ORDER BY/SORT BY/DISTRIBUTE BY/CLUSTER BY.
CREATE_TABLE_STAGING_LOCATION
Create a catalog table in a staging location.
DESC_TABLE_COLUMN_PARTITION
DESC TABLE COLUMN for a specific partition.
DROP_DATABASE
Drop the default database <database>
.
DROP_NAMESPACE
Drop the namespace <namespace>
.
DROP_NAMESPACE_CASCADE
Drop namespace cascade is not supported in <provider>
catalogs.
HIVE_TABLE_TYPE
The <tableName>
is hive <tableType>
.
HIVE_WITH_ANSI_INTERVALS
Hive table <tableName>
with ANSI intervals.
INSERT_PARTITION_SPEC_IF_NOT_EXISTS
INSERT INTO <tableName>
with IF NOT EXISTS in the PARTITION spec.
LATERAL_COLUMN_ALIAS_IN_AGGREGATE_FUNC
Referencing a lateral column alias <lca>
in the aggregate function <aggFunc>
.
LATERAL_COLUMN_ALIAS_IN_AGGREGATE_WITH_WINDOW_AND_HAVING
Referencing lateral column alias <lca>
in the aggregate query both with window expressions and with having clause. Please rewrite the aggregate query by removing the having clause or removing lateral alias reference in the SELECT list.
LATERAL_COLUMN_ALIAS_IN_GROUP_BY
Referencing a lateral column alias via GROUP BY alias/ALL is not supported yet.
LATERAL_COLUMN_ALIAS_IN_WINDOW
Referencing a lateral column alias <lca>
in window expression <windowExpr>
.
LATERAL_JOIN_USING
JOIN USING with LATERAL correlation.
LITERAL_TYPE
Literal for ‘<value>
’ of <type>
.
MULTI_ACTION_ALTER
The target JDBC server hosting table <tableName>
does not support ALTER TABLE with multiple actions. Split the ALTER TABLE up into individual actions to avoid this error.
ORC_TYPE_CAST
Unable to convert <orcType>
of Orc to data type <toType>
.
OVERWRITE_BY_SUBQUERY
INSERT OVERWRITE with a subquery condition.
PANDAS_UDAF_IN_PIVOT
Pandas user defined aggregate function in the PIVOT clause.
PARAMETER_MARKER_IN_UNEXPECTED_STATEMENT
Parameter markers are not allowed in <statement>
.
PARTITION_WITH_NESTED_COLUMN_IS_UNSUPPORTED
Invalid partitioning: <cols>
is missing or is in a map or array.
PIVOT_AFTER_GROUP_BY
PIVOT clause following a GROUP BY clause. Consider pushing the GROUP BY into a subquery.
PIVOT_TYPE
Pivoting by the value ‘<value>
’ of the column data type <type>
.
PURGE_PARTITION
Partition purge.
PYTHON_DATA_SOURCE
Python data sources.
PYTHON_UDF_IN_ON_CLAUSE
Python UDF in the ON clause of a <joinType>
JOIN. In case of an INNER JOIN consider rewriting to a CROSS JOIN with a WHERE clause.
REPLACE_NESTED_COLUMN
The replace function does not support nested column <colName>
.
SET_NAMESPACE_PROPERTY
<property>
is a reserved namespace property, <msg>
.
SET_OPERATION_ON_MAP_TYPE
Cannot have MAP
type columns in DataFrame which calls set operations (INTERSECT, EXCEPT, etc.), but the type of column <colName>
is <dataType>
.
SET_PROPERTIES_AND_DBPROPERTIES
set PROPERTIES and DBPROPERTIES at the same time.
SET_TABLE_PROPERTY
<property>
is a reserved table property, <msg>
.
SET_VARIABLE_USING_SET
<variableName>
is a VARIABLE and cannot be updated using the SET statement. Use SET VARIABLE <variableName>
= … instead.
TABLE_OPERATION
Table <tableName>
does not support <operation>
. Please check the current catalog and namespace to make sure the qualified table name is expected, and also check the catalog implementation which is configured by “spark.sql.catalog”.
TEMP_VARIABLE_ON_DBSQL
Temporary variables are not yet supported for queries from SQL warehouses.
TIME_TRAVEL
Time travel on the relation: <relationId>
.
TOO_MANY_TYPE_ARGUMENTS_FOR_UDF_CLASS
UDF class with <num>
type arguments.