Updated on 2023-10-23 GMT+08:00

Statistics Information Functions

Statistics information functions are divided into the following two categories: functions that access databases, using the OID of each table or index in a database to mark the database for which statistics are generated; functions that access servers, identified by the server process ID, whose value ranges from 1 to the number of currently active servers.

  • pg_stat_get_db_conflict_tablespace(oid)

    Description: Specifies the number of queries canceled due to a conflict between the restored tablespace and the deleted tablespace in the database.

    Return type: bigint

  • pg_control_group_config

    Description: Prints Cgroup configurations on the current node.

    Return type: record

  • pg_stat_get_db_stat_reset_time(oid)

    Description: Specifies the most recent time when database statistics were reset. It is initialized to the system time during the first connection to each database. The reset time is updated when you call pg_stat_reset on the database and execute pg_stat_reset_single_table_counters against any table or index in it.

    Return type: timestamptz

  • pg_stat_get_function_total_time(oid)

    Description: Specifies the total wall clock time spent in the function, in microseconds. The time spent on this function calling other functions is included.

    Return type: bigint

  • pg_stat_get_xact_tuples_returned(oid)

    Description: Specifies the number of rows read through sequential scans when the parameter is a table in the current transaction or the number of index entries returned when the parameter is an index.

    Return type: bigint

  • pg_lock_status()

    Description: Queries information about locks held by open transactions. All users can execute this function.

    Return type: For details, see PG_LOCKS which is obtained by querying this function.

  • pg_stat_get_xact_numscans(oid)

    Description: Specifies the number of sequential scans performed when the parameter is a table in the current transaction or the number of index scans performed when the parameter is an index.

    Return type: bigint

  • pg_stat_get_xact_blocks_fetched(oid)

    Description: Specifies the number of disk block fetch requests for a table or an index in the current transaction.

    Return type: bigint

  • pg_stat_get_xact_blocks_hit(oid)

    Description: Specifies the number of disk block fetch requests for tables or indexes found in cache in the current transaction.

    Return type: bigint

  • pg_stat_get_xact_function_calls(oid)

    Description: Specifies the number of times the function is called in the current transaction.

    Return type: bigint

  • pg_stat_get_xact_function_self_time(oid)

    Description: Specifies the time spent on this function in the current transaction, excluding the time spent on this function internally calling other functions.

    Return type: bigint

  • pg_stat_get_xact_function_total_time(oid)

    Description: Specifies the total wall clock time (in microseconds) spent on the function in the current transaction, including the time spent on this function internally calling other functions.

    Return type: bigint

  • pg_stat_get_wal_senders()

    Description: Queries walsender information on the primary node.

    Return type: setofrecord

    The following table describes return columns.

    Table 1 Return column description

    Column

    Type

    Description

    pid

    bigint

    Thread ID of the WAL sender

    sender_pid

    integer

    Lightweight thread ID of the WAL sender

    local_role

    text

    Type of the primary node

    peer_role

    text

    Type of the standby node

    peer_state

    text

    Status of the standby node

    state

    text

    Status of the WAL sender

    catchup_start

    timestamp with time zone

    Startup time of a catchup task

    catchup_end

    timestamp with time zone

    End time of a catchup task

    sender_sent_location

    text

    Sending position of the primary node

    sender_write_location

    text

    Writing position of the primary node

    sender_flush_location

    text

    Flushing position of the primary node

    sender_replay_location

    text

    Redo position of the primary node

    receiver_received_location

    text

    Receiving position of the standby node

    receiver_write_location

    text

    Writing position of the standby node

    receiver_flush_location

    text

    Flushing position of the standby node

    receiver_replay_location

    text

    Redo position of the standby node

    sync_percent

    text

    Synchronization percentage

    sync_state

    text

    Synchronization status

    sync_group

    text

    Group to which the synchronous replication belongs

    sync_priority

    text

    Priority of synchronous replication

    sync_most_available

    text

    Maximum availability mode

    channel

    text

    Channel information of the WAL sender

  • get_paxos_replication_info()

    Description: Queries the primary/standby replication status in Paxos mode.

    Return type: setofrecord

    The following table describes return columns.

    Table 2 Return column description

    Column

    Type

    Description

    paxos_write_location

    text

    Location of the Xlog that has been written to the Distribute Consensus Framework (DCF)

    paxos_commit_location

    text

    Location of the Xlog agreed in the DCF

    local_write_location

    text

    Writing position of a node

    local_flush_location

    text

    Flushing position of a node

    local_replay_location

    text

    Redo position of a node

    dcf_replication_info

    text

    DCF module information of a node

  • pg_stat_get_stream_replications()

    Description: Queries the primary/standby replication status.

    Return type: setofrecord

    The following table describes return values.

    Table 3 Return value description

    Return Parameter

    Type

    Description

    local_role

    text

    Local role

    static_connections

    integer

    Connection statistics

    db_state

    text

    Database status

    detail_information

    text

    Detailed information

  • pg_stat_get_db_numbackends(oid)

    Description: Specifies the number of active server processes for a database.

    Return type: integer

  • pg_stat_get_db_xact_commit(oid)

    Description: Specifies the number of transactions committed in a database.

    Return type: bigint

  • pg_stat_get_db_xact_rollback(oid)

    Description: Specifies the number of transactions rolled back in a database.

    Return type: bigint

  • pg_stat_get_db_blocks_fetched(oid)

    Description: Specifies the number of disk blocks fetch requests for a database.

    Return type: bigint

  • pg_stat_get_db_blocks_hit(oid)

    Description: Specifies the number of disk block fetch requests found in cache for a database.

    Return type: bigint

  • pg_stat_get_db_tuples_returned(oid)

    Description: Specifies the number of tuples returned for a database.

    Return type: bigint

  • pg_stat_get_db_tuples_fetched(oid)

    Description: Specifies the number of tuples fetched for a database.

    Return type: bigint

  • pg_stat_get_db_tuples_inserted(oid)

    Description: Specifies the number of tuples inserted in a database.

    Return type: bigint

  • pg_stat_get_db_tuples_updated(oid)

    Description: Specifies the number of tuples updated in a database.

    Return type: bigint

  • pg_stat_get_db_tuples_deleted(oid)

    Description: Specifies the number of tuples deleted in a database.

    Return type: bigint

  • pg_stat_get_db_conflict_lock(oid)

    Description: Specifies the number of lock conflicts in a database.

    Return type: bigint

  • pg_stat_get_db_deadlocks(oid)

    Description: Specifies the number of deadlocks in a database.

    Return type: bigint

  • pg_stat_get_numscans(oid)

    Description: Specifies the number of sequential row scans done if parameters are in a table or the number of index scans done if parameters are in an index.

    Return type: bigint

  • pg_stat_get_role_name(oid)

    Description: Obtains the username based on the user OID. Only users with the sysadmin or monitor admin permission can access the information.

    Return type: text

    Example:

    openGauss=# select pg_stat_get_role_name(10);
     pg_stat_get_role_name
    -----------------------
     aabbcc
    (1 row)
  • pg_stat_get_tuples_returned(oid)

    Description: Specifies the number of sequential row scans done if parameters are in a table or the number of index scans done if parameters are in an index.

    Return type: bigint

  • pg_stat_get_tuples_fetched(oid)

    Description: Specifies the number of table rows fetched by bitmap scans if parameters are in a table or the number of table rows fetched by simple index scans using the index if parameters are in an index.

    Return type: bigint

  • pg_stat_get_tuples_inserted(oid)

    Description: Specifies the number of rows inserted into a table.

    Return type: bigint

  • pg_stat_get_tuples_updated(oid)

    Description: Specifies the number of rows updated in a table.

    Return type: bigint

  • pg_stat_get_tuples_deleted(oid)

    Description: Specifies the number of rows deleted from a table.

    Return type: bigint

  • pg_stat_get_tuples_changed(oid)

    Description: Specifies the total number of inserted, updated, and deleted rows after a table was last analyzed or autoanalyzed.

    Return type: bigint

  • pg_stat_get_tuples_hot_updated(oid)

    Description: Specifies the number of rows hot updated in a table.

    Return type: bigint

  • pg_stat_get_live_tuples(oid)

    Description: Specifies the number of live rows in a table.

    Return type: bigint

  • pg_stat_get_dead_tuples(oid)

    Description: Specifies the number of dead rows in a table.

    Return type: bigint

  • pg_stat_get_blocks_fetched(oid)

    Description: Specifies the number of disk block fetch requests for a table or an index.

    Return type: bigint

  • pg_stat_get_blocks_hit(oid)

    Description: Specifies the number of disk block requests found in cache for a table or an index.

    Return type: bigint

  • pg_stat_get_partition_tuples_inserted(oid)

    Description: Specifies the number of rows in the corresponding table partition.

    Return type: bigint

  • pg_stat_get_partition_tuples_updated(oid)

    Description: Specifies the number of rows that have been updated in the corresponding table partition.

    Return type: bigint

  • pg_stat_get_partition_tuples_deleted(oid)

    Description: Specifies the number of rows deleted from the corresponding table partition.

    Return type: bigint

  • pg_stat_get_partition_tuples_changed(oid)

    Description: Specifies the total number of inserted, updated, and deleted rows after a table partition was last analyzed or autoanalyzed.

    Return type: bigint

  • pg_stat_get_partition_live_tuples(oid)

    Description: Specifies the number of live rows in a partitioned table.

    Return type: bigint

  • pg_stat_get_partition_dead_tuples(oid)

    Description: Specifies the number of dead rows in a partitioned table.

    Return type: bigint

  • pg_stat_get_xact_tuples_fetched(oid)

    Description: Specifies the number of tuple rows scanned in a transaction.

    Return type: bigint

  • pg_stat_get_xact_tuples_inserted(oid)

    Description: Specifies the number of tuple inserted into the active subtransactions related to a table.

    Return type: bigint

  • pg_stat_get_xact_tuples_deleted(oid)

    Description: Specifies the number of deleted tuples in the active subtransactions related to a table.

    Return type: bigint

  • pg_stat_get_xact_tuples_hot_updated(oid)

    Description: Specifies the number of hot updated tuples in the active subtransactions related to a table.

    Return type: bigint

  • pg_stat_get_xact_tuples_updated(oid)

    Description: Specifies the number of updated tuples in the active subtransactions related to a table.

    Return type: bigint

  • pg_stat_get_xact_partition_tuples_inserted(oid)

    Description: Specifies the number of inserted tuples in the active subtransactions related to a table partition.

    Return type: bigint

  • pg_stat_get_xact_partition_tuples_deleted(oid)

    Description: Specifies the number of deleted tuples in the active subtransactions related to a table partition.

    Return type: bigint

  • pg_stat_get_xact_partition_tuples_hot_updated(oid)

    Description: Specifies the number of hot updated tuples in the active subtransactions related to a table partition.

    Return type: bigint

  • pg_stat_get_xact_partition_tuples_updated(oid)

    Description: Specifies the number of updated tuples in the active subtransactions related to a table partition.

    Return type: bigint

  • pg_stat_get_last_vacuum_time(oid)

    Description: Specifies the most recent time when the autovacuum thread is manually started to clear a table.

    Return type: timestamptz

  • pg_stat_get_last_autovacuum_time(oid)

    Description: Specifies the time of the last vacuum initiated by the autovacuum daemon on a table.

    Return type: timestamptz

  • pg_stat_get_vacuum_count(oid)

    Description: Specifies the number of times a table is manually cleared.

    Return type: bigint

  • pg_stat_get_autovacuum_count(oid)

    Description: Specifies the number of times the autovacuum daemon is started to clear a table.

    Return type: bigint

  • pg_stat_get_last_analyze_time(oid)

    Description: Specifies the last time when a table starts to be analyzed manually or by the autovacuum thread.

    Return type: timestamptz

  • pg_stat_get_last_autoanalyze_time(oid)

    Description: Specifies the time when the last analysis initiated by the autovacuum daemon on a table.

    Return type: timestamptz

  • pg_stat_get_analyze_count(oid)

    Description: Specifies the number of times a table is manually analyzed.

    Return type: bigint

  • pg_stat_get_autoanalyze_count(oid)

    Description: Specifies the number of times the autovacuum daemon analyzes a table.

    Return type: bigint

  • pg_total_autovac_tuples(bool,bool)

    Description: Returns tuple records related to the total autovac, such as nodename, nspname, relname, and tuple IUDs. The input parameters specify whether to query relation and local information, respectively.

    Return type: setofrecord

    The following table describes return parameters.

    Table 4 Return parameter description

    Return Parameter

    Type

    Description

    nodename

    name

    Node name

    nspname

    name

    Name of a namespace

    relname

    name

    Name of an object, such as a table, an index, or a view

    partname

    name

    Partition name

    n_dead_tuples

    bigint

    Number of dead rows in a table partition

    n_live_tuples

    bigint

    Number of live rows in a table partition

    changes_since_analyze

    bigint

    Number of changes generated by ANALYZE

  • pg_autovac_status(oid)

    Description: Returns autovac information, such as nodename, nspname, relname, analyze, vacuum, thresholds of analyze and vacuum, and the number of analyzed or vacuumed tuples. Only users with the sysadmin permission can use this function.

    Return type: setofrecord

    The following table describes return parameters.

    Table 5 Return parameter description

    Return Parameter

    Type

    Description

    nspname

    text

    Name of a namespace

    relname

    text

    Name of an object, such as a table, an index, or a view

    nodename

    text

    Node name

    doanalyze

    Boolean

    Whether to execute ANALYZE

    anltuples

    bigint

    Number of ANALYZE tuples

    anlthresh

    bigint

    ANALYZE threshold

    dovacuum

    Boolean

    Whether to execute VACUUM

    vactuples

    bigint

    Number of VACUUM tuples

    vacthresh

    bigint

    VACUUM threshold

  • pg_autovac_timeout(oid)

    Description: Returns the number of consecutive timeouts during the autovac operation on a table. If the table information is invalid or the node information is abnormal, NULL will be returned.

    Return type: bigint

  • pg_stat_get_last_data_changed_time(oid)

    Description: Returns the time when INSERT, UPDATE, DELETE, or EXCHANGE/TRUNCATE/DROP PARTITION was last performed on a table. The data in the last_data_changed column of the PG_STAT_ALL_TABLES view is calculated by using this function. The performance of obtaining the last modification time by using the view is poor when the table has a large amount of data. In this case, you are advised to use the function.

    Return type: timestamptz

  • pg_stat_set_last_data_changed_time(oid)

    Description: Manually changes the time when INSERT, UPDATE, DELETE, or EXCHANGE/TRUNCATE/DROP PARTITION was last performed.

    Return type: void

  • pg_backend_pid()

    Description: Specifies the thread ID of the server thread attached to the current session.

    Return type: integer

  • pg_stat_get_activity(integer)

    Description: Returns a record about the backend with the specified PID. A record for each active backend in the system is returned if NULL is specified. The returned result does not contain the connection_info column. The initial user, system administrators and users with the monadmin permission can view all data. Common users can only query their own results.

    Example:

    openGauss=# select * from pg_stat_get_activity(139881386280704);
     datid |       pid       | sessionid | usesysid | application_name | state  |                        query                         | waiting |          xact_start           |          query_start          |
        backend_start         |         state_change         | client_addr | client_hostname | client_port | enqueue |     query_id      |   srespool   | global_sessionid | unique_sql_id | trace_id
    -------+-----------------+-----------+----------+------------------+--------+------------------------------------------------------+---------+-------------------------------+-------------------------------+-----
    --------------------------+------------------------------+-------------+-----------------+-------------+---------+-------------------+--------------+------------------+---------------+----------
     16545 | 139881386280704 |        69 |       10 | gsql             | active | select * from pg_stat_get_activity(139881386280704); | f       | 2022-01-18 19:43:05.167718+08 | 2022-01-18 19:43:05.167718+08 | 2022
    -01-18 19:42:33.513507+08 | 2022-01-18 19:43:05.16773+08 |             |                 |          -1 |         | 72620543991624410 | default_pool | 1938253334#69#0  |    3751941862 |
    (1 row)

    Return type: setofrecord

    The following table describes return parameters.

    Table 6 Return parameter description

    Return Parameter

    Type

    Description

    datid

    oid

    OID of the database that the user session connects to in the backend

    pid

    bigint

    Backend thread ID

    sessionid

    bigint

    Session ID

    usesysid

    oid

    OID of the user logged in to the backend

    application_name

    text

    Name of the application connected to the backend

    state

    text

    Overall status of the backend

    query

    text

    Latest query at the backend. If state is active, this column shows the ongoing query. In all other states, it shows the last query that was executed.

    waiting

    Boolean

    Whether the backend is currently waiting on a lock. If yes, the value is true.

    xact_start

    timestamp with time zone

    Time when current transaction was started (null if no transaction is active).

    If the current query is the first of its transaction, the value of this column is the same as that of the query_start column.

    query_start

    timestamp with time zone

    Time when the currently active query was started, or time when the last query was started if state is not active

    backend_start

    timestamp with time zone

    Time when this process was started, that is, when the client connected to the server

    state_change

    timestamp with time zone

    Time when state was last modified

    client_addr

    inet

    IP address of the client connected to the backend. If this column is NULL, it indicates either the client is connected via a Unix socket on the server or this is an internal process, such as AUTOVACUUM.

    client_hostname

    text

    Host name of the connected client, as reported by a reverse DNS lookup of client_addr. This column will be non-null only for IP connections and only when log_hostname is enabled.

    client_port

    integer

    TCP port number that the client uses for communication with this backend (–1 if a Unix socket is used)

    enqueue

    text

    Unsupported currently

    query_id

    bigint

    ID of a query

    srespool

    name

    Name of the resource pool

    global_sessionid

    text

    Global session ID

    unique_sql_id

    bigint

    Unique SQL statement ID

    trace_id

    text

    Driver-specific trace ID, which is associated with an application request

  • pg_stat_get_activity_with_conninfo(integer)

    Description: Returns a record about the backend with the specified PID. A record for each active backend in the system is returned if NULL is specified. The initial user, system administrators and users with the monadmin permission can view all data. Common users can only query their own results.

    Return type: setofrecord

    The following table describes return values.

    Table 7 Return value description

    Return Value

    Return Type

    Description

    datid

    oid

    OID of the database that the user session connects to in the backend

    pid

    bigint

    Backend thread ID

    sessionid

    bigint

    Session ID

    usesysid

    oid

    OID of the user logged in to the backend

    application_name

    text

    Name of the application connected to the backend

    state

    text

    Overall status of the backend

    query

    text

    Latest query at the backend. If state is active, this column shows the ongoing query. In all other states, it shows the last query that was executed.

    waiting

    Boolean

    Whether the backend is currently waiting on a lock. If yes, the value is true.

    xact_start

    timestamp with time zone

    Time when current transaction was started (null if no transaction is active). If the current query is the first of its transaction, the value of this column is the same as that of the query_start column.

    query_start

    timestamp with time zone

    Time when the currently active query was started, or time when the last query was started if state is not active

    backend_start

    timestamp with time zone

    Time when this process was started, that is, when the client connected to the server

    state_change

    timestamp with time zone

    Time when state was last modified

    client_addr

    inet

    IP address of the client connected to the backend If this column is NULL, it indicates either the client is connected via a Unix socket on the server or this is an internal process, such as AUTOVACUUM.

    client_hostname

    text

    Host name of the connected client, as reported by a reverse DNS lookup of client_addr. This column will be non-null only for IP connections and only when log_hostname is enabled.

    client_port

    integer

    TCP port number that the client uses for communication with this backend (–1 if a Unix socket is used)

    enqueue

    text

    Unsupported currently

    query_id

    bigint

    ID of a query

    connection_info

    text

    A string in JSON format recording the driver type, driver version, driver deployment path, and process owner of the connected database

    srespool

    name

    Name of the resource pool

    global_sessionid

    text

    Global session ID

    unique_sql_id

    bigint

    Unique SQL statement ID

    trace_id

    text

    Driver-specific trace ID, which is associated with an application request

  • pg_user_iostat(text)

    Description: Displays the I/O load management information about the job currently executed by the user. (The current feature is a lab feature. Contact Huawei technical support before using it.)

    Return type: record

    The following table describes return fields.

    Name

    Type

    Description

    userid

    oid

    User ID

    min_curr_iops

    int4

    Minimum I/O of the current user across database nodes. The IOPS is counted by ones for column storage and by ten thousands for row storage.

    max_curr_iops

    int4

    Maximum I/O of the current user across database nodes. The IOPS is counted by ones for column storage and by ten thousands for row storage.

    min_peak_iops

    int4

    Minimum peak I/O of the current user across database nodes. The IOPS is counted by ones for column storage and by ten thousands for row storage.

    max_peak_iops

    int4

    Maximum peak I/O of the current user across database nodes. The IOPS is counted by ones for column storage and by ten thousands for row storage.

    io_limits

    int4

    io_limits set for the resource pool specified by the user. The IOPS is counted by ones for column storage and by ten thousands for row storage.

    io_priority

    text

    io_priority set for the user. The IOPS is counted by ones for column storage and by ten thousands for row storage.

    curr_io_limits

    int4

    Real-time io_limits value when io_priority is used to control I/Os

  • pg_stat_get_function_calls(oid)

    Description: Specifies the number of times the function has been called.

    Return type: bigint

  • pg_stat_get_function_self_time(oid)

    Description: Specifies the time spent in only this function. The time spent on this function calling other functions is excluded.

    Return type: bigint

  • pg_stat_get_backend_idset()

    Description: Sets the number of currently active server processes (from 1 to the number of active server processes).

    Return type: setofinteger

  • pg_stat_get_backend_pid(integer)

    Description: Specifies the ID of the given server thread.

    Return type: bigint

  • pg_stat_get_backend_dbid(integer)

    Description: Specifies the ID of the database connected to the given server process.

    Return type: oid

  • pg_stat_get_backend_userid(integer)

    Description: Specifies the user ID of the given server process.

    Return type: oid

  • pg_stat_get_backend_activity(integer)

    Description: Active command of the given server process, but only if the current user is a system administrator or the same user as that of the session being queried and track_activities is on

    Return type: text

  • pg_stat_get_backend_waiting(integer)

    Description: True if the given server process is waiting for a lock, but only if the current user is a system administrator or the same user as that of the session being queried and track_activities is on

    Return type: Boolean

  • pg_stat_get_backend_activity_start(integer)

    Description: Specifies the time when the given server process's currently executing query is started only if the current user is the system administrator or the user of the session being queried and track_activities is enabled.

    Return type: timestamp with time zone

  • pg_stat_get_backend_xact_start(integer)

    Description: Specifies the time when the given server process's currently executing transaction is started only if the current user is the system administrator or the user of the session being queried and track_activities is enabled.

    Return type: timestamp with time zone

  • pg_stat_get_backend_start(integer)

    Description: Specifies the time when the given server process is started. If the current user is neither the system administrator nor the user of the session being queried, NULL is returned.

    Return type: timestamp with time zone

  • pg_stat_get_backend_client_addr(integer)

    Description: Specifies the IP address of the client connected to the given server process. If the connection is over a Unix domain socket, or if the current user is neither a system administrator nor the same user as that of the session being queried, NULL will be returned.

    Return type: inet

  • pg_stat_get_backend_client_port(integer)

    Description: Specifies the TCP port number of the client connected to the given server process. If the connection is over a Unix domain socket, -1 will be returned. If the current user is neither a system administrator nor the same user as that of the session being queried, NULL will be returned.

    Return type: integer

  • pg_stat_get_bgwriter_timed_checkpoints()

    Description: Specifies the time when the background writer starts scheduled checkpoints (because the checkpoint_timeout time has expired).

    Return type: bigint

  • pg_stat_get_bgwriter_requested_checkpoints()

    Description: Specifies the time when the background writer starts checkpoints based on requests from the backend because checkpoint_segments has been exceeded or the CHECKPOINT command has been executed.

    Return type: bigint

  • pg_stat_get_bgwriter_buf_written_checkpoints()

    Description: Specifies the number of buffers written by the background writer during checkpoints.

    Return type: bigint

  • pg_stat_get_bgwriter_buf_written_clean()

    Description: Specifies the number of buffers written by the background writer for routine cleaning of dirty pages.

    Return type: bigint

  • pg_stat_get_bgwriter_maxwritten_clean()

    Description: Specifies the time when the background writer stops its cleaning scan because it has written more buffers than specified in the bgwriter_lru_maxpages parameter.

    Return type: bigint

  • pg_stat_get_buf_written_backend()

    Description: Specifies the number of buffers written by the backend because they need to allocate a new buffer.

    Return type: bigint

  • pg_stat_get_buf_alloc()

    Description: Specifies the total number of the allocated buffers.

    Return type: bigint

  • pg_stat_clear_snapshot()

    Description: Discards the current statistics snapshot.

    Return type: void

  • pg_stat_reset()

    Description: Resets all statistics counters for the current database to zero (requires system administrator permissions).

    Return type: void

  • pg_stat_reset_shared(text)

    Description: Resets all statistics counters for the current database in each node in a shared cluster to zero (requires system administrator permissions).

    Return type: void

  • pg_stat_reset_single_table_counters(oid)

    Description: Resets statistics for a single table or index in the current database to zero (requires system administrator permissions).

    Return type: void

  • pg_stat_reset_single_function_counters(oid)

    Description: Resets statistics for a single function in the current database to zero (requires system administrator permissions).

    Return type: void

  • pg_stat_session_cu(int, int, int)

    Description: Obtains the compression unit (CU) hit statistics of sessions running on the current node.

    Return type: record

  • pg_stat_get_cu_mem_hit(oid)

    Description: Obtains the number of CU memory hits of a column storage table in the current database of the current node.

    Return type: bigint

  • pg_stat_get_cu_hdd_sync(oid)

    Description: Obtains the times CU is synchronously read from a disk by a column storage table in the current database of the current node.

    Return type: bigint

  • pg_stat_get_cu_hdd_asyn(oid)

    Description: Obtains the times CU is asynchronously read from a disk by a column storage table in the current database of the current node.

    Return type: bigint

  • pg_stat_get_db_cu_mem_hit(oid)

    Description: Obtains the CU memory hit in a database of the current node.

    Return type: bigint

  • pg_stat_get_db_cu_hdd_sync(oid)

    Description: Obtains the times CU is synchronously read from a disk by a database of the current node.

    Return type: bigint

  • fenced_udf_process(integer)

    Description: Shows the number of local UDF Master and Work processes. If the input parameter is set to 1, the number of Master processes is queried. If the input parameter is set to 2, the number of Worker processes is queried. If the input parameter is set to 3, all Worker processes are killed.

    Return type: text

  • total_cpu()

    Description: Obtains the CPU time used by the current node, in jiffies.

    Return type: bigint

  • total_memory()

    Description: Obtains the size of the virtual memory used by the current node, in KB.

    Return type: bigint

  • pg_stat_get_db_cu_hdd_asyn(oid)

    Description: Obtains the times CU is asynchronously read from a disk by a database of the current node.

    Return type: bigint

  • pg_stat_bad_block(text, int, int, int, int, int, timestamp with time zone, timestamp with time zone)

    Description: Obtains damage information about pages or CUs after the current node is started.

    Example: select * from pg_stat_bad_block();

    Return type: record

  • pg_stat_bad_block_clear()

    Description: Deletes the page and CU damage information that is read and recorded on the node (requires system administrator permissions).

    Return type: void

  • gs_respool_exception_info(pool text)

    Description: Queries the query rule of a specified resource pool.

    Return type: record

  • gs_control_group_info(pool text)

    Description: Queries information about Cgroups associated with a resource pool.

    Return type: record

    The command output is as follows:

    Attribute

    Value

    Description

    name

    class_a:workload_a1

    Class name and workload name

    class

    class_a

    Class Cgroup name

    workload

    workload_a1

    Workload Cgroup name

    type

    DEFWD

    Cgroup type (Top, CLASS, BAKWD, DEFWD, and TSWD)

    gid

    87

    Cgroup ID

    shares

    30

    Percentage of CPU resources to those on the parent node

    limits

    0

    Percentage of CPU cores to those on the parent node

    rate

    0

    Allocation ratio in Timeshare

    cpucores

    0-3

    Number of CPU cores

  • gs_all_control_group_info()

    Description: Collects information about all Cgroups in the database.

    Return type: record

  • gs_get_control_group_info()

    Description: Collects information about all Cgroups.

    Return type: record

  • get_instr_workload_info(integer)

    Description: Obtains the transaction volume and time information on the primary database node.

    Return type: record

    Attribute

    Value

    Description

    resourcepool_oid

    10

    OID of the resource pool (the logic is equivalent to the load)

    commit_counter

    4

    Number of frontend transactions that were committed

    rollback_counter

    1

    Number of frontend transactions that were rolled back

    resp_min

    949

    Minimum response time of frontend transactions (unit: μs)

    resp_max

    201891

    Maximum response time of frontend transactions (unit: μs)

    resp_avg

    43564

    Average response time of frontend transactions (unit: μs)

    resp_total

    217822

    Total response time of frontend transactions (unit: μs)

    bg_commit_counter

    910

    Number of backend transactions that were committed

    bg_rollback_counter

    0

    Number of backend transactions that were rolled back

    bg_resp_min

    97

    Minimum response time of backend transactions (unit: μs)

    bg_resp_max

    678080687

    Maximum response time of backend transactions (unit: μs)

    bg_resp_avg

    327847884

    Average response time of backend transactions (unit: μs)

    bg_resp_total

    298341575300

    Total response time of backend transactions (unit: μs)

  • pv_instance_time()

    Description: Obtains the time consumed in each execution phase on the current node.

    Return type: record

    Stat_name Attribute

    Value

    Description

    DB_TIME

    1062385

    Total end-to-end wall time consumed by all threads (unit: μs)

    CPU_TIME

    311777

    Total CPU time consumed by all threads (unit: μs)

    EXECUTION_TIME

    380037

    Total time consumed on the executor (unit: μs)

    PARSE_TIME

    6033

    Total time consumed for parsing SQL statements (unit: μs)

    PLAN_TIME

    173356

    Total time consumed for generating an execution plan (unit: μs)

    REWRITE_TIME

    2274

    Total time consumed on query rewriting (unit: μs)

    PL_EXECUTION_TIME

    0

    Total time consumed for executing PL/SQL statements (unit: μs)

    PL_COMPILATION_TIME

    557

    Total time consumed for SQL compilation (unit: μs)

    NET_SEND_TIME

    1673

    Total time consumed for sending data over network (unit: μs)

    DATA_IO_TIME

    426622

    Total time consumed for data read and write (unit: μs)

  • DBE_PERF.get_global_instance_time()

    Description: Provides the time consumed in each key phase in the entire database. To query this function, you must have the sysadmin permission.

    Return type: record

  • get_instr_unique_sql()

    Description: Obtains information about execution statements (normalized SQL statements) on the current node as a user with the sysadmin permission.

    Return type: record

  • reset_unique_sql(text, text, bigint)

    Description: Resets information about system execution statements (normalized SQL statements) information as a user with the sysadmin permission. The value of the first parameter can be global or local. global indicates that information on all nodes is cleared, and local indicates that only information on the current node is cleared. The value of the second parameter can be ALL, BY_USERID, or BY_CNID. ALL indicates that all information is cleared. BY_USERID indicates that the SQL information of the user specified by USERID is cleared. BY_CNID indicates that the SQL information related to the primary node of the database in the system is cleared. The third parameter indicates CNID and USERID. If the second parameter is set to ALL, the third parameter does not take effect and can be set to any value.

    Return type: Boolean

    This function involves distributed nodes. Currently, GaussDB is a centralized database, for which the function of the value global is the same as that of the value local and the second parameter cannot set to be BY_CNID.

  • get_instr_wait_event(NULL)

    Description: Obtains the statistics on wait events of the current node.

    Return type: record

  • get_instr_user_login()

    Description: Obtains the number of user login and logout times on the current node. Only users with the sysadmin or monitor admin permission can execute this function.

    Return type: record

  • get_instr_rt_percentile(integer)

    Description: Obtains the SQL response time P80 and P95 distribution information of the database.

    Return type: record

  • get_node_stat_reset_time()

    Description: Obtains statistics about reset (restart, primary/standby switchover, and database deletion) time of the current node.

    Return type: record

  • DBE_PERF.get_global_os_runtime()

    Description: Displays the running status of the current operating system. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_os_threads()

    Description: Provides information about the threads under all normal nodes of the entire database. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_workload_sql_count()

    Description: Provides statistics about the number of SELECT, UPDATE, INSERT, DELETE, DDL, DML, and DCL statements of different service loads in the entire database. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_workload_sql_elapse_time()

    Description: Provides statistics about the number of SELECT, UPDATE, INSERT, and DELETE statements and response time information (TOTAL, AVG, MIN, and MAX) for different loads in the entire database. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_workload_transaction()

    Description: Obtains the transaction volume and time information on all nodes of the database. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_session_stat()

    Description: Obtains the session status information on all nodes of the database. To query this function, you must have the sysadmin permission.

    Return type: record

    The status information contains the following 17 items: commit, rollback, sql, table_scan, blocks_fetched, physical_read_operation, shared_blocks_dirtied, local_blocks_dirtied, shared_blocks_read, local_blocks_read, blocks_read_time, blocks_write_time, sort_imemory, sort_idisk, cu_mem_hit, cu_hdd_sync_read, and cu_hdd_asyread.

  • DBE_PERF.get_global_session_time()

    Description: Provides the time consumed in each key phase of each node in the entire database. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_session_memory()

    Description: Displays statistics about memory usage at the session level on each node in the unit of MB, including all the memory allocated to Postgres and stream threads on DNs for jobs currently executed by users. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_session_memory_detail()

    Description: Displays statistics about thread memory usage on each node by MemoryContext node. To query this function, you must have the sysadmin permission.

    Return type: record

  • create_wlm_session_info(int flag)

    Description: Clears top SQL query statement-level statistics recorded in the current memory. Only the administrator can execute this function.

    Return type: int

  • pg_stat_get_wlm_session_info(int flag)

    Description: Obtains top SQL query statement-level statistics recorded in the current memory. If the input parameter is not 0, the information is cleared from the memory. Only users with the system admin or monitor admin permission can execute this function.

    Return type: record

  • gs_paxos_stat_replication()

    Description: Queries the standby node information on the primary node. Currently, only the centralized DCF mode is supported.

    Return type: setofrecord

    The following table describes return columns.

    Column

    Type

    Description

    local_role

    text

    Role of the node that sends logs

    peer_role

    text

    Role of the node that receives logs

    local_dcf_role

    text

    DCF role of the node that sends logs

    peer_dcf_role

    text

    DCF role of the node that receives logs

    peer_state

    text

    Status of the node that receives logs

    sender_write_location

    text

    Location in the Xlog buffer where the node that sends logs is written

    sender_commit_location

    text

    Consistency point reached for the DCF logs of the node that sends logs

    sender_flush_location

    text

    Location in the Xlog disk where the node that sends logs is written

    sender_replay_location

    text

    Location where the node that sends logs replays logs

    receiver_write_location

    text

    Location in the Xlog buffer where the node that receives logs is written

    receiver_commit_location

    text

    Consistency point reached for the DCF logs of the node that receives logs

    receiver_flush_location

    text

    Location in the Xlog disk where the node that receives logs is written

    receiver_replay_location

    text

    Location where the node that receives logs replays Xlogs

    sync_percent

    text

    Synchronization percentage

    dcf_run_mode

    int4

    DCF synchronization mode

    channel

    text

    Channel information

  • gs_wlm_get_resource_pool_info(int)

    Description: Obtains resource usage statistics of all users. The input parameter can be any value of the INT type or be NULL.

    Return type: record

  • gs_wlm_get_all_user_resource_info()

    Description: Obtains resource usage statistics of all users.

    Return type: record

  • gs_wlm_get_user_info(int)

    Description: Obtains information about all users. The input parameter is of the int type and can be any int value or NULL. Only users with the sysadmin permission can execute this function.

    Return type: record

  • gs_wlm_get_workload_records()

    Description: Obtains all job information in dynamic load management. This function is valid only when dynamic load management is enabled. (The current feature is a lab feature. Contact Huawei technical support before using it.)

    Return type: record

  • gs_wlm_readjust_user_space()

    Description: Corrects the storage space usage of all users. Only the administrator can execute this function.

    Return type: record

  • gs_wlm_readjust_user_space_through_username(text name)

    Description: Corrects the storage space usage of a specified user. Common users can use this function to modify only their own usage. Only the administrator can modify the usage of all users. If the value of name is 0000, the usage of all users needs to be modified.

    Return type: record

  • gs_wlm_readjust_user_space_with_reset_flag(text name, boolean isfirst)

    Description: Corrects the storage space usage of a specified user. If the input parameter isfirst is set to true, statistics are collected from 0. Otherwise, statistics are collected from the previous result. Common users can use this function to modify only their own usage. Only the administrator can modify the usage of all users. If the value of name is 0000, the usage of all users needs to be modified.

    Return type: record

  • gs_wlm_session_respool(bigint)

    Description: Obtains the session resource pool information about all backend threads. The input parameter can be any value of the bigint type or can be null.

    Return type: record

  • gs_wlm_get_session_info()

    Description: This API has been discarded and is unavailable currently.

  • gs_wlm_get_user_session_info()

    Description: This API has been discarded and is unavailable currently.

  • gs_io_wait_status()

    Description: This API does not support single-node systems or centralized systems and is unavailable currently.

  • global_stat_get_hotkeys_info()

    Description: Obtains the statistics on hot keys in the entire database instance. This API does not support single-node systems or centralized systems and is unavailable currently.

  • global_stat_clean_hotkeys()

    Description: Clears statistics on hot keys in the entire database instance. This API does not support single-node systems or centralized systems and is unavailable currently.

  • DBE_PERF.get_global_session_stat_activity()

    Description: Displays information about threads that are running on each node in the database. To query this function, you must have the monitoradmin permission.

    Return type: record

  • DBE_PERF.get_global_thread_wait_status()

    Description: Displays the block waiting status of backend threads and auxiliary threads on all nodes. To query this function, you must have the sysadmin or monitoradmin permission.

    Return type: record

  • DBE_PERF.get_global_operator_history_table()

    Description: Displays the operator-related records (persistent) generated after jobs are executed on the primary database node of the current user. To query this function, you must have the sysadmin or monitoradmin permission.

    Return type: record

  • DBE_PERF.get_global_operator_history()

    Description: Displays the operator-related records generated after jobs are executed on the primary database node of the current user. To query this function, you must have the sysadmin or monitoradmin permission.

    Return type: record

  • DBE_PERF.get_global_operator_runtime()

    Description: Displays real-time operator-related records of jobs executed on the primary database node of the current user. To query this function, you must have the sysadmin or monitoradmin permission.

    Return type: record

  • DBE_PERF.get_global_statement_complex_history()

    Description: Displays the historical records of complex queries on the primary database node of the current user. To query this function, you must have the monitoradmin permission.

    Return type: record

  • DBE_PERF.get_global_statement_complex_history_table()

    Description: Displays the historical records (persistent) of complex queries on the primary database node of the current user. To query this function, you must have the monitoradmin permission.

    Return type: record

  • DBE_PERF.get_global_statement_complex_runtime()

    Description: Displays the real-time information of complex queries on the primary database node of the current user. To query this function, you must have the sysadmin or monitoradmin permission.

    Return type: record

  • DBE_PERF.get_global_memory_node_detail()

    Description: Displays the memory usage of a certain database on all nodes. To query this function, you must have the monitoradmin permission.

    Return type: record

  • DBE_PERF.get_global_shared_memory_detail()

    Description: Displays the usage information about all the shared memory contexts of all nodes. To query this function, you must have the monitoradmin permission.

    Return type: record

  • DBE_PERF.get_global_statio_all_indexes()

    Description: Displays statistics about each index displayed in a row in the current database, showing I/O statistics about accesses to that specific index. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_stat_all_tables()

    Description: Displays statistics about a row in each table (including the TOAST table) on each node.

    Return type: record

  • DBE_PERF.get_global_stat_all_tables()

    Description: Displays statistics about a row in each table (including the TOAST table) on each node.

    Return type: record

  • DBE_PERF.get_local_toastname_and_toastindexname()

    Description: Provides the mapping between the name and index of the local TOAST table and its associated table. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_statio_all_indexes()

    Description: Collects statistics about each index displayed in a row in the current databases of all nodes and displays the I/O statistics of a specific index. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_statio_all_sequences()

    Description: Provides I/O status information about all sequences in the namespace. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_statio_all_tables()

    Description: Displays the I/O statistics about each table in databases on each node. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_statio_all_tables()

    Description: Collects I/O statistics about each table in databases in the database. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_local_toast_relation()

    Description: Provides the mapping between the name of the local TOAST table and its associated table. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_statio_sys_indexes()

    Description: Displays the I/O status information about all system catalog indexes in namespaces on each node. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_statio_sys_indexes()

    Description: Collects the I/O status information about all system catalog indexes in namespaces on each node. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_statio_sys_sequences()

    Description: Provides the I/O status information about all the system sequences in the namespace. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_statio_sys_tables()

    Description: Provides I/O status information about all system catalogs in namespaces on each node. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_statio_sys_tables()

    Description: Displays the I/O status information of all system catalogs in the namespace in the database. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_statio_user_indexes()

    Description: Displays the I/O status information about all user relationship table indexes in namespaces on each node. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_statio_user_indexes()

    Description: Displays the I/O status information about all user relationship table indexes in namespaces in the database. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_statio_user_sequences()

    Description: Displays the I/O status information about all user sequences in the namespace of each node. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_statio_user_tables()

    Description: Displays the I/O status information about all user relationship tables in namespaces on each node. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_statio_user_tables()

    Description: Displays the I/O status information about all user relationship tables in namespaces in the database. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_stat_db_cu()

    Description: Queries CU hits in a database and in each node in the database. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_stat_all_indexes()

    Description: Displays statistics of each index in databases on all nodes. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_stat_all_indexes()

    Description: Collects statistics of each index in all databases on all nodes. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_stat_sys_tables()

    Description: Displays statistics about the system catalogs of all the namespaces in pg_catalog and information_schema schemas on each node. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_stat_sys_tables()

    Description: Collects statistics about the system catalogs of all the namespaces in pg_catalog and information_schema schemas on each node. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_stat_sys_indexes()

    Description: Displays index status information about all the system catalogs in the pg_catalog and information_schema schemas on each node. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_stat_sys_indexes()

    Description: Collects statistics about index status information about all the system catalogs in the pg_catalog and information_schema schemas on each node. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_stat_user_tables()

    Description: Displays the status information about customized ordinary tables in all namespaces. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_stat_user_tables()

    Description: Collects statistics about the status information about customized ordinary tables in all namespaces. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_stat_user_indexes()

    Description: Displays the status information about the index of customized ordinary tables in all databases. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_stat_user_indexes()

    Description: Collects statistics about the status information about the index of customized ordinary tables in all databases. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_stat_database()

    Description: Displays database statistics of all nodes. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_stat_database_conflicts()

    Description: Collects statistics on the database of all nodes. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_stat_xact_all_tables()

    Description: Displays transaction status information about all ordinary tables and TOAST tables in all namespaces. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_stat_xact_all_tables()

    Description: Collects statistics about transaction status information about all ordinary tables and TOAST tables in all namespaces. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_stat_xact_sys_tables()

    Description: Displays transaction status information about all system catalogs in namespaces on all nodes. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_stat_xact_sys_tables()

    Description: Collects statistics about transaction status information about all system catalogs in namespaces on all nodes. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_stat_xact_user_tables()

    Description: Displays the transaction status information of the user tables in the namespaces on all nodes. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_stat_xact_user_tables()

    Description: Collects statistics about the transaction status information of the user tables in the namespaces on all nodes. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_stat_user_functions()

    Description: Displays the transaction status information of customized functions in the namespaces on all nodes. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_stat_xact_user_functions()

    Description: Collects statistics about the transaction status information of customized functions in the namespaces on all nodes. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_stat_bad_block()

    Description: Displays information about table and index read failures on all nodes. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_file_redo_iostat()

    Description: Collects statistics on information about table and index read failures on all nodes. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_file_iostat()

    Description: Displays statistics about data file I/Os on all nodes. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_locks()

    Description: Displays lock information of all nodes. To query this function, you must have the sysadmin or monadmin permission.

    Return type: record

  • DBE_PERF.get_global_replication_slots()

    Description: Displays logical replication information on all nodes. To query this function, you must have the sysadmin or monadmin permission.

    Return type: record

  • DBE_PERF.get_global_bgwriter_stat()

    Description: Displays statistics about the background writer process's activities on all nodes. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_replication_stat()

    Description: Displays information about log synchronization status on each node, such as the locations where the sender sends logs and where the receiver receives logs. To query this function, you must have the sysadmin or monadmin permission.

    Return type: record

  • DBE_PERF.get_global_transactions_running_xacts()

    Description: Displays information about running transactions on each node. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_transactions_running_xacts()

    Description: Collects statistics of information about running transactions on each node. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_transactions_prepared_xacts()

    Description: Displays information about transactions that are currently prepared for two-phase commit on each node. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_transactions_prepared_xacts()

    Description: Collects statistics information about transactions that are currently prepared for two-phase commit on each node. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_statement()

    Description: Displays the status information of the historically-executed statements on each node. To query this function, you must have the sysadmin or monitor admin permission.

    Return type: record

  • DBE_PERF.get_global_statement_count()

    Description: Displays the number of SELECT, UPDATE, INSERT, and DELETE statements and response time information (TOTAL, AVG, MIN, and MAX) on each node. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_config_settings()

    Description: Displays GUC parameter configuration information on each node. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_wait_events()

    Description: Displays the wait event status information on each node. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_statement_responsetime_percentile()

    Description: Obtains the response time distribution for 80% and 95% SQL statements of the database. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_summary_user_login()

    Description: Collects statistics about number of user login and logout times on each node in the database. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.get_global_record_reset_time()

    Description: Displays the statistics about reset (restart, primary/standby switchover, and database deletion) time of the database. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.track_memory_context(context_list text)

    Description: Sets the memory context whose memory application details need to be collected. The input parameter is the memory context names, which are separated by commas (,), for example, ThreadTopMemoryContext, SessionCacheMemoryContext. Note that the memory context names are context-sensitive. In addition, the length of a single memory context is 63, and the excess part is truncated. The maximum number of memory contexts that can be collected at a time is 16. If the number of memory contexts exceeds 16, the setting fails. Each time this function is called, the previous statistics result is cleared. When the input parameter is set to "", the statistics function is disabled. Only the initial user or a user with the monadmin permission can execute this function.

    Return type: Boolean

  • DBE_PERF.track_memory_context_detail()

    Description: Obtains the memory application details of the memory context specified by the DBE_PERF.track_memory_context function. For details, see the DBE_PERF.track_memory_context_detail view. Only the initial user or a user with the monadmin permission can execute this function.

    Return type: record

  • pg_stat_get_mem_mbytes_reserved(tid)

    Description: Collects statistics on variables related to resource management, which is used only for fault locating.

    Parameter: thread ID

    Return type: text

  • gs_wlm_user_resource_info(name text)

    Description: Queries a user's resource quota and resource usage.

    Return type: record

  • pg_stat_get_file_stat()

    Description: Records statistics about data file I/Os to indicate I/O performance and detect performance problems such as abnormal I/O operations.

    Return type: record

  • pg_stat_get_redo_stat()

    Description: Displays statistics on the replay of session thread logs.

    Return type: record

  • pg_stat_get_status(int8)

    Description: Tests the block waiting status about the backend thread and auxiliary thread of the current instance.

    Return type: record

  • get_local_rel_iostat()

    Description: Queries the accumulated I/O status of data files on the current node.

    Return type: record

  • DBE_PERF.get_global_rel_iostat()

    Description: Displays statistics about data file I/Os on all nodes. To query this function, you must have the sysadmin permission.

    Return type: record

  • DBE_PERF.global_threadpool_status()

    Description: Displays the status of worker threads and sessions in thread pools on all nodes. For details about the columns returned by the function, see GLOBAL_THREADPOOL_STATUS.

    Return type: record

  • remote_bgwriter_stat()

    Description: Displays the information about pages flushed by the bgwriter threads of all instances in the database, number of pages in the candidate buffer chain, and buffer elimination information (except for the local node and not available on the DN).

    Return type: record

  • pv_os_run_info

    Description: Displays the running status of the current OS. For details about the columns, see GS_OS_RUN_INFO.

    Parameter: nan

    Return type: SETOF record

  • pv_session_stat

    Description: Collects session status information by session thread or AutoVacuum thread. For details about the columns, see GS_SESSION_STAT.

    Parameter: nan

    Return type: SETOF record

  • pv_session_time

    Description: Collects statistics on the running time of session threads and the time consumed in each execution phase. For details about the columns, see GS_SESSION_TIME.

    Parameter: nan

    Return type: SETOF record

  • pg_stat_get_db_temp_bytes

    Description: Collects statistics on the total amount of data written to temporary files through database query. All temporary files are counted, regardless of why the temporary file was created, and regardless of the log_temp_files setting.

    Parameter: oid

    Return type: bigint

  • pg_stat_get_db_temp_files

    Description: Queries the number of temporary files created in the database. All temporary files are counted, regardless of why the temporary file was created (for example, sorting or hashing), and regardless of the log_temp_files setting.

    Parameter: oid

    Return type: bigint

  • create_wlm_instance_statistics_info

    Description: Saves the historical monitoring data of the current instance persistently.

    Parameter: nan

    Return type: integer

    Table 8 remote_bgwriter_stat parameter description

    Parameter

    Type

    Description

    node_name

    text

    Instance name

    bgwr_actual_flush_total_num

    bigint

    Total number of dirty pages flushed by the bgwriter thread from the startup time to the current time

    bgwr_last_flush_num

    integer

    Number of dirty pages flushed by the bgwriter thread in the previous batch

    candidate_slots

    integer

    Number of pages in the current candidate buffer chain

    get_buffer_from_list

    bigint

    Number of times that pages are obtained from the candidate buffer chain during buffer eviction

    get_buf_clock_sweep

    bigint

    Number of times that pages are obtained from the original eviction solution during buffer eviction

  • remote_candidate_stat()

    Description: Displays the checkpoint information and log flushing information about all instances in the database (except the current node). Centralized systems are not supported.

    Return type: record

  • remote_ckpt_stat()

    Description: Displays the checkpoint information and log flushing information about all instances in the database (except the current node). Centralized systems are not supported.

    Return type: record

  • remote_single_flush_dw_stat()

    Description: Displays the single-page doublewrite file status of all instances in the database (except the current node). Centralized systems are not supported.

    Return type: record

  • remote_double_write_stat()

    Description: Displays doublewrite file status of all instances in the database (except the current node). Centralized systems are not supported.

    Return type: record

  • remote_pagewriter_stat()

    Description: Displays the page flushing information and checkpoint information about all instances in the database (except the current node). Centralized systems are not supported.

    Return type: record

  • remote_recovery_status()

    Description: Displays log flow control information about the primary and standby nodes (except the current node). Centralized systems are not supported.

    Return type: record

  • remote_redo_stat()

    Description: Displays the log replay status of all instances in the database (except the current node). Centralized systems are not supported.

    Return type: record

    Example:

    The function pg_backend_pid shows the ID of the current server thread.

    1
    2
    3
    4
    5
    openGauss=# SELECT pg_backend_pid();
     pg_backend_pid  
    -----------------
     139706243217168
    (1 row)
    

    The function pg_stat_get_backend_pid shows the ID of a given server thread.

    1
    2
    3
    4
    5
    openGauss=# SELECT pg_stat_get_backend_pid(1);
     pg_stat_get_backend_pid 
    -------------------------
             139706243217168
    (1 row)
    
  • dbe_perf.gs_stat_activity_timeout(int)

    Description: Obtains information about query jobs whose execution time exceeds the timeout threshold on the current node. The correct result can be returned only when the GUC parameter track_activities is set to on. The timeout threshold ranges from 0 to 2147483.

    Return type: SETOF record

    Name

    Type

    Description

    database

    name

    Name of the database to which a user session is connected

    pid

    bigint

    Backend thread ID

    sessionid

    bigint

    Session ID

    usesysid

    oid

    OID of the user logged in to the backend

    application_name

    text

    Name of the application connected to the backend

    query

    text

    Query that is being executed on the backend

    xact_start

    timestamptz

    Time when the current transaction is started

    query_start

    timestamptz

    Time when the current query starts

    query_id

    bigint

    Query statement ID

  • gs_wlm_user_resource_info(name text)

    Description: Queries a user's resource quota and resource usage. Common users can query only their own information. Administrators can query information about all users.

    Return type: record

  • create_wlm_instance_statistics_info

    Description: Saves the historical monitoring data of the current instance persistently.

    Parameter: nan

    Return type: integer

  • gs_session_memory

    Description: Collects statistics about memory usage at the session level, including all the memory allocated to Postgres and Stream threads on DNs for tasks currently executed by users.

    If enable_memory_limit is set to off, this function cannot be used.

    Return type: record

    Table 9 Return value description

    Name

    Type

    Description

    sessid

    text

    Thread start time and ID

    init_mem

    integer

    Memory allocated to the currently executed jobs before they enter the executor, in MB

    used_mem

    integer

    Memory allocated to the currently executed jobs, in MB

    peak_mem

    integer

    Peak memory allocated to the currently executed jobs, in MB

  • gs_wlm_persistent_user_resource_info()

    Description: Archives all user resource usage statistics to the gs_wlm_user_resource_history system catalog. To query this function, you must have the sysadmin permission.

    Return type: record

  • create_wlm_operator_info(int flag)

    Description: Clears top SQL operator-level statistics recorded in the current memory. If the input parameter is greater than 0, the information is archived to gs_wlm_operator_info and gs_wlm_ec_operator_info. Otherwise, the information is not archived. Only users with the sysadmin permission can execute this function.

    Return type: int

  • GS_ALL_NODEGROUP_CONTROL_GROUP_INFO(text)
    Description: Provides Cgroup information for all logical database instances. Before calling this function, you need to specify the name of the logical database instance to be queried. For example, to query the Cgroup information for the installation logical database instance, run the following command:
    1
    SELECT * FROM GS_ALL_NODEGROUP_CONTROL_GROUP_INFO('installation')
    

    Return type: record

    The following table describes return columns.

    Name

    Type

    Description

    name

    text

    Cgroup name.

    type

    text

    Cgroup type.

    gid

    bigint

    Cgroup ID.

    classgid

    bigint

    ID of the class Cgroup where a workload Cgroup belongs.

    class

    text

    Class Cgroup.

    workload

    text

    Workload Cgroup.

    shares

    bigint

    CPU quota allocated to the Cgroup.

    limits

    bigint

    Limit of CPU resources allocated to a Cgroup.

    wdlevel

    bigint

    Workload Cgroup level.

    cpucores

    text

    Information about the CPU cores used by a Cgroup.

  • gs_total_nodegroup_memory_detail

    Description: Returns information about the memory used by the current logical database, in MB.

    Return type: SETOF record

  • local_redo_time_count()

    Description: Returns the time consumption statistics on each process of each playback thread on the current node (valid data exists only on the standby node).

    The return values are as follows:

    local_redo_time_count parameters

    Column

    Description

    thread_name

    Thread name

    step1_total

    Total duration of step 1. The process of each thread is as follows:

    Ultimate RTO

    • batch redo: obtains a log from a queue.
    • redo manager: obtains a log from a queue.
    • redo worker: obtains a log from a queue.
    • trxn manager: reads a log from a queue.
    • trxn worker: reads a log from a queue.
    • read worker: reads an Xlog page (overall) from a file.
    • read page worker: obtains a log from a queue.
    • startup: obtains a log from a queue.

    Parallel replay:

    • page redo: obtains a log from a queue.
    • startup: reads a log.

    step1_count

    Number of accumulated execution times of step 1.

    step2_total

    Total duration of step 2. The process of each thread is as follows:

    Ultimate RTO

    • batch redo: processes logs (overall).
    • redo manager: processes logs (overall).
    • redo worker: processes logs (overall).
    • trxn manager: processes logs (overall).
    • trxn worker: processes logs (overall).
    • redo worker: specifies the time required for reading the Xlog page.
    • read page worker: generates and sends LSN forwarders.
    • startup: checks whether to replay to the specified position.

    Parallel replay:

    • page redo: processes logs (overall).
    • startup: checks whether to replay to the specified position.

    step2_count

    Number of accumulated execution times of step 2.

    step3_total

    Total duration of step 3. The process of each thread is as follows:

    Ultimate RTO

    • batch redo: updates the standby state.
    • redo manager: processes data logs.
    • redo worker: replays page logs (overall).
    • trxn manager: updates the flush LSN.
    • trxn worker: replays logs.
    • redo worker: pushes the Xlog segment.
    • read page worker: obtains a new item.
    • startup: collects statistics on the wait time of delayed replay feature.

    Parallel replay:

    • page redo: updates the standby state.
    • startup: collects statistics on the wait time of delayed replay feature.

    step3_count

    Number of accumulated execution times of step 3.

    step4_total

    Total duration of step 4. The process of each thread is as follows:

    Ultimate RTO:

    • batch redo: parses Xlogs.
    • redo manager: processes DDL.
    • redo worker: reads data pages.
    • trxn manager: synchronizes the wait time.
    • trxn worker: updates the LSN of the current thread.
    • read page worker: stores logs in the distribution thread.
    • startup: distributes logs (overall).

    Parallel replay:

    • page redo: replays undo logs.
    • startup: distributes logs (overall).

    step4_count

    Number of accumulated execution times of step 4.

    step5_total

    Total duration of step 5. The process of each thread is as follows:

    Ultimate RTO:

    • batch redo: distributes logs to the redo manager.
    • redo manager: distributes logs to redo workers.
    • redo worker: replays data page logs.
    • trxn manager: distributes data to the trxn worker.
    • trxn worker: forcibly synchronizes the wait time.
    • read page worker: updates the LSN of the current thread.
    • startup: decodes logs.

    Parallel replay:

    • page redo: replays sharetrxn logs.
    • startup: replays logs.

    step5_count

    Number of accumulated execution times of step 5.

    step6_total

    Total duration of step 6. The process of each thread is as follows:

    Ultimate RTO:

    • redo worker: replays non-data page logs.
    • trxn manager: updates global LSNs.
    • read page worker: performs log CRC check.

    Parallel replay:

    • page redo: replays synctrxn logs.
    • startup: forcibly synchronizes the wait time.

    step6_count

    Number of accumulated execution times of step 6.

    step7_total

    Total duration of step 7. The process of each thread is as follows:

    Ultimate RTO:

    • redo manager: creates tablespaces.
    • redo worker: updates FSM.

    Parallel replay:

    page redo: replays a single log.

    step7_count

    Number of accumulated execution times of step 7.

    step8_total

    Total duration of step 8. The process of each thread is as follows:

    Ultimate RTO:

    redo worker: forcibly synchronizes the wait time.

    Parallel replay:

    page redo: replays all workers do log.

    step8_count

    Number of accumulated execution times of step 8.

    step9_total

    Total duration of step 9. The process of each thread is as follows:

    Ultimate RTO:

    None

    Parallel replay:

    page redo: replays muliti workers do log.

    step9_count

    Number of accumulated execution times of step 9.

  • local_xlog_redo_statics()

    Description: Returns the statistics on each type of logs that have been replayed on the current node (valid data exists only on the standby node).

    The return values are as follows:
    Table 10 local_xlog_redo_statics parameters

    Column

    Description

    xlog_type

    Log types.

    rmid

    resource manager id

    info

    xlog operation

    num

    Number of logs.

    extra

    Valid values are available for page replay logs and xact logs. The page replay log indicates the number of pages read from the disk. The xact log indicates the number of deleted files.

  • gs_get_shared_memctx_detail(text)

    Description: Returns the memory application details of the specified memory context, including the file, line number, and size of each memory application (the size of the same line in the same file is accumulated). Only the memory context queried through the pg_shared_memory_detail view can be queried. The input parameter is the memory context name (that is, the contextname column in the result returned by the pg_shared_memory_detail view). To query this function, you must have the sysadmin or monitor admin permission.

    Return type: SETOF record

    Name

    Type

    Description

    file

    text

    Name of the file where the memory is applied for.

    line

    int8

    Line number of the code in the file where the requested memory is located.

    size

    int8

    Size of the applied memory. The value is accumulated if the memory is applied for multiple times in the same line of the same file.

    This view is not supported in the Lite release version.

  • gs_get_session_memctx_detail(text)

    Description: Returns the memory application details of the specified memory context, including the file, line number, and size of each memory application (the size of the same line in the same file is accumulated). This parameter is valid only in thread pool mode. Only the memory context queried through the gs_session_memory_context view can be queried. The input parameter is the memory context name (that is, the contextname column in the result returned by the gs_session_memory_context view). To query this function, you must have the sysadmin or monitor admin permission.

    Return type: SETOF record

    Name

    Type

    Description

    file

    text

    Name of the file where the memory is applied for.

    line

    int8

    Line number of the code in the file where the requested memory is located.

    size

    int8

    Size of the allocated memory, in bytes. The value is accumulated if the memory is allocated for multiple times to the same line of the same file.

    This view takes effect only in thread pool mode and is not supported in the Lite release version.

  • gs_get_history_memory_detail(cstring)

    Description: Queries historical memory snapshot information. The input parameter type is cstring. The value can be NULL or the name of the memory snapshot log file.

    1. If the value of the input parameter is NULL, the list of all memory snapshot log files on the current node is displayed.
    2. If the value of the input parameter is the name of the memory snapshot log file in the list queried in 1, the detailed information about the memory snapshot recorded in the log file is displayed.
    3. If you enter any other input parameter, the system displays a message indicating that the input parameter is incorrect or the file fails to be opened.

    To query this function, you must have the sysadmin or monitor admin permission.

    Return type: text

    Name

    Type

    Description

    memory_info

    text

    Memory information. If the input parameter of the function is set to NULL, the memory snapshot file list is displayed. If the input parameter is set to the name of the memory snapshot file, the content of the file is displayed.

  • gs_stack()

    Description: Displays the call stack of a thread. To query this function, you must have the sysadmin or monadmin permission.

    Parameter: tid, which indicates the thread ID. tid is an optional parameter. If it is specified, the function returns the call stack of the thread corresponding to tid. If it is not specified, the function returns the call stacks of all threads.

    Return value: If tid is specified, the return value is of the TEXT type. If tid is not specified, the return value is a SETOF record.

    Example:

     1
     2
     3
     4
     5
     6
     7
     8
     9
    10
    11
    12
    13
    openGauss=# SELECT gs_stack(139663481165568);
                                  gs_stack
    --------------------------------------------------------------------
     __poll + 0x2d                                                     +
     WaitLatchOrSocket(Latch volatile*, int, int, long) + 0x29f        +
     WaitLatch(Latch volatile*, int, long) + 0x2e                      +
     JobScheduleMain() + 0x90f                                         +
     int GaussDbThreadMain<(knl_thread_role)9>(knl_thread_arg*) + 0x456+
     InternalThreadFunc(void*) + 0x2d                                  +
     ThreadStarterFunc(void*) + 0xa4                                   +
     start_thread + 0xc5                                               +
     clone + 0x6d                                                      +
    (1 row)
    

  • gs_get_thread_memctx_detail(tid,text)

    Description: Returns the memory application details of the specified memory context, including the file, line number, and size of each memory application (the size of the same line in the same file is accumulated). Only the memory context queried through the gs_thread_memory_context view is supported. The first input parameter is the thread ID (the tid column of the data returned by the gs_thread_memory_context), and the second parameter is the memory context name (the contextname column of the data returned by gs_thread_memory_context). To query this function, you must have the sysadmin or monitor admin permission.

    Return type: SETOF record

    Name

    Type

    Description

    file

    text

    Name of the file where the memory is applied for.

    line

    int8

    Line number of the code in the file where the requested memory is located.

    size

    int8

    Size of the allocated memory, in bytes. The value is accumulated if the memory is allocated for multiple times to the same line of the same file.

    This view is not supported in the Lite release version.