Updated on 2024-04-30 GMT+08:00

PG_INDEX

PG_INDEX records part of the information about indexes. The rest is mostly in PG_CLASS.

Table 1 PG_INDEX columns

Name

Type

Description

indexrelid

oid

OID of the pg_class entry for this index

indrelid

oid

OID of the pg_class entry for the table this index is for

indnatts

smallint

Number of columns in an index

indisunique

boolean

This index is a unique index if the value is true.

indisprimary

boolean

This index represents the primary key of the table if the value is true. If this value is true, the value of indisunique is true.

indisexclusion

boolean

This index supports exclusion constraints if the value is true.

indimmediate

boolean

A uniqueness check is performed upon data insertion if the value is true.

indisclustered

boolean

The table was last clustered on this index if the value is true.

indisusable

boolean

This index supports insert/select if the value is true.

indisvalid

boolean

This index is valid for queries if the value is true. If this column is false, this index is possibly incomplete and must still be modified by INSERT/UPDATE operations, but it cannot safely be used for queries. If it is a unique index, the uniqueness property is also not true.

indcheckxmin

boolean

If the value is true, queries must not use the index until the xmin of this row in pg_index is below their TransactionXmin event horizon, because the table may contain broken HOT chains with incompatible rows that they can see.

indisready

boolean

If the value is true, this index is ready for inserts. If the value is false, this index is ignored when data is inserted or modified.

indkey

int2vector

This is an array of indnatts values that indicate which table columns this index creates. For example, a value of 1 3 means that the first and the third columns make up the index key. 0 in this array indicates that the corresponding index attribute is an expression over the table columns, rather than a simple column reference.

indcollation

oidvector

ID of each column used by the index

indclass

oidvector

For each column in the index key, this column contains the OID of the operator class to use. For details, see PG_OPCLASS.

indoption

int2vector

Array of values that store per-column flag bits. The meaning of the bits is defined by the index's access method.

indexprs

pg_node_tree

Expression trees (in nodeToString() representation) for index attributes that are not simple column references. It is a list with one element for each zero entry in INDKEY. NULL if all index attributes are simple references.

indpred

pg_node_tree

Expression tree (in nodeToString() representation) for partial index predicate. If the index is not a partial index, the value is null.

indnullstreatment

tinyint

Processing mode of the NULL value in the unique index. This field is valid only if indisunique is set to true.

Options:

  • 0: NULLS DISTINCT. NULL values are not equivalent and can be inserted repeatedly.
  • 1: NULLS NOT DISTINCT. NULL values are equivalent and cannot be inserted repeatedly.
  • 2: NULLS IGNORE. NULL columns are ignored during equivalent comparison. If all index columns are NULL, NULL values can be inserted repeatedly. If part of the index columns are NULL, data can be inserted only if non-null values are different.

Default value: 0

NOTE:
  • If the current cluster was upgraded from an earlier version to 8.2.0.100, the value of this field is NULL for existing indexes. For newly created indexes, the value of this field is determined by the [ NULLS [NOT] DISTINCT | NULLS IGNORE ] field. The default value is 0.
  • If the current cluster is newly installed and its version is 8.2.0.100, for newly created indexes, the value of this field is determined by the [ NULLS [NOT] DISTINCT | NULLS IGNORE ] field. The default value is 0.