The gp_persistent_relation_node table table keeps track of the status of file system objects in relation to the transaction status of relation objects (tables, view, indexes, and so on). This information is used to make sure the state of the system catalogs and the file system files persisted to disk are synchronized. This information is used by the primary to mirror file replication process.

Table 1. pg_catalog.gp_persistent_relation_node
column type references description
tablespace_oid oid pg_tablespace.oid Tablespace object id
database_oid oid pg_database.oid Database object id
relfilenode_oid oid pg_class.relfilenode The object id of the relation file node.
segment_file_num integer   For append-optimized tables, the append-optimized segment file number.
relation_storage_manager smallint   Whether the relation is heap storage or append-optimized storage.
persistent_state smallint   0 - free

1 - create pending

2 - created

3 - drop pending

4 - aborting create

5 - "Just in Time" create pending

6 - bulk load create pending

mirror_existence_state smallint   0 - none

1 - not mirrored

2 - mirror create pending

3 - mirrorcreated

4 - mirror down before create

5 - mirror down during create

6 - mirror drop pending

7 - only mirror drop remains

parent_xid integer   Global transaction id.
persistent_serial_num bigint   Log sequence number position in the transaction log for a file block.
previous_free_tid tid   Used by Greenplum Database to internally manage persistent representations of file system objects.