v3-0001-Add-pg_stat_recovery-system-view.patch
application/x-patch
Filename: v3-0001-Add-pg_stat_recovery-system-view.patch
Type: application/x-patch
Part: 0
Message:
Re: Add pg_stat_recovery system view
Patch
Format: format-patch
Series: patch v3-0001
Subject: Add pg_stat_recovery system view
| File | + | − |
|---|---|---|
| doc/src/sgml/monitoring.sgml | 150 | 0 |
| src/backend/access/transam/xlogfuncs.c | 98 | 0 |
| src/backend/catalog/system_views.sql | 13 | 0 |
| src/include/catalog/pg_proc.dat | 7 | 0 |
| src/test/regress/expected/rules.out | 10 | 0 |
| src/test/regress/expected/sysviews.out | 7 | 0 |
| src/test/regress/sql/sysviews.sql | 3 | 0 |
From f23f2662f41d8456a1450c98e4dd1b50ea43ed99 Mon Sep 17 00:00:00 2001
From: alterego655 <824662526@qq.com>
Date: Tue, 27 Jan 2026 12:11:26 +0800
Subject: [PATCH v3 1/3] Add pg_stat_recovery system view
Introduce pg_stat_recovery to expose WAL recovery state maintained by the
startup process, including:
- last replayed record boundaries and timeline
- current replay end pointer and timeline
- current WAL chunk start time
- promotion trigger state
- recovery pause state
Implement pg_stat_get_recovery() as a non-SRF, following the
pg_stat_get_wal_receiver() pattern.
Gate all output behind pg_read_all_stats. If the caller lacks that role,
or recovery is not in progress, return NULL immediately before taking
XLogRecoveryCtl->info_lck.
---
doc/src/sgml/monitoring.sgml | 150 +++++++++++++++++++++++++
src/backend/access/transam/xlogfuncs.c | 98 ++++++++++++++++
src/backend/catalog/system_views.sql | 13 +++
src/include/catalog/pg_proc.dat | 7 ++
src/test/regress/expected/rules.out | 10 ++
src/test/regress/expected/sysviews.out | 7 ++
src/test/regress/sql/sysviews.sql | 3 +
7 files changed, 288 insertions(+)
diff --git a/doc/src/sgml/monitoring.sgml b/doc/src/sgml/monitoring.sgml
index dcf6e6a2f48..3bace7d1430 100644
--- a/doc/src/sgml/monitoring.sgml
+++ b/doc/src/sgml/monitoring.sgml
@@ -338,6 +338,15 @@ postgres 27093 0.0 0.0 30096 2752 ? Ss 11:34 0:00 postgres: ser
</entry>
</row>
+ <row>
+ <entry><structname>pg_stat_recovery</structname><indexterm><primary>pg_stat_recovery</primary></indexterm></entry>
+ <entry>Only one row, showing statistics about the recovery state of the
+ startup process. This view returns no row when not in recovery.
+ See <link linkend="monitoring-pg-stat-recovery-view">
+ <structname>pg_stat_recovery</structname></link> for details.
+ </entry>
+ </row>
+
<row>
<entry><structname>pg_stat_recovery_prefetch</structname><indexterm><primary>pg_stat_recovery_prefetch</primary></indexterm></entry>
<entry>Only one row, showing statistics about blocks prefetched during recovery.
@@ -1912,6 +1921,147 @@ description | Waiting for a newly initialized WAL file to reach durable storage
</sect2>
+ <sect2 id="monitoring-pg-stat-recovery-view">
+ <title><structname>pg_stat_recovery</structname></title>
+
+ <indexterm>
+ <primary>pg_stat_recovery</primary>
+ </indexterm>
+
+ <para>
+ The <structname>pg_stat_recovery</structname> view will contain only
+ one row, showing statistics about the recovery state of the startup
+ process. This view returns no row when the server is not in recovery.
+ </para>
+
+ <para>
+ Some columns are restricted to members of the
+ <literal>pg_read_all_stats</literal> role; other users will see
+ <literal>NULL</literal> values for these columns. The restricted columns
+ include LSN values, timeline IDs, and timestamps. Basic operational status
+ columns (<structfield>promote_triggered</structfield>,
+ <structfield>pause_state</structfield>) are visible to all users.
+ </para>
+
+ <table id="pg-stat-recovery-view" xreflabel="pg_stat_recovery">
+ <title><structname>pg_stat_recovery</structname> View</title>
+ <tgroup cols="1">
+ <thead>
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ Column Type
+ </para>
+ <para>
+ Description
+ </para></entry>
+ </row>
+ </thead>
+
+ <tbody>
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>promote_triggered</structfield> <type>boolean</type>
+ </para>
+ <para>
+ True if a promotion has been triggered.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>last_replayed_read_lsn</structfield> <type>pg_lsn</type>
+ </para>
+ <para>
+ Start write-ahead log location of the last successfully replayed
+ WAL record.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>last_replayed_end_lsn</structfield> <type>pg_lsn</type>
+ </para>
+ <para>
+ End write-ahead log location of the last successfully replayed
+ WAL record.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>last_replayed_tli</structfield> <type>integer</type>
+ </para>
+ <para>
+ Timeline of the last successfully replayed WAL record.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>replay_end_lsn</structfield> <type>pg_lsn</type>
+ </para>
+ <para>
+ Write-ahead log location of the record currently being replayed
+ (end position plus one). When no record is being actively replayed,
+ equals <structfield>last_replayed_end_lsn</structfield>.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>replay_end_tli</structfield> <type>integer</type>
+ </para>
+ <para>
+ Timeline of the WAL record currently being replayed.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>current_chunk_start_time</structfield> <type>timestamp with time zone</type>
+ </para>
+ <para>
+ Time when the startup process observed that replay had caught up
+ with the latest received WAL chunk. Used in recovery-conflict
+ timing and replay/apply-lag diagnostics. NULL if not yet
+ available.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>pause_state</structfield> <type>text</type>
+ </para>
+ <para>
+ Recovery pause state. Possible values are:
+ </para>
+ <itemizedlist>
+ <listitem>
+ <para>
+ <literal>not paused</literal>: Recovery is proceeding normally.
+ </para>
+ </listitem>
+ <listitem>
+ <para>
+ <literal>pause requested</literal>: A pause has been requested
+ but recovery has not yet paused.
+ </para>
+ </listitem>
+ <listitem>
+ <para>
+ <literal>paused</literal>: Recovery is paused.
+ </para>
+ </listitem>
+ </itemizedlist>
+ </entry>
+ </row>
+
+ </tbody>
+ </tgroup>
+ </table>
+
+ </sect2>
+
<sect2 id="monitoring-pg-stat-recovery-prefetch">
<title><structname>pg_stat_recovery_prefetch</structname></title>
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 2efe4105efb..142289d8071 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -22,10 +22,13 @@
#include "access/xlog_internal.h"
#include "access/xlogbackup.h"
#include "access/xlogrecovery.h"
+#include "catalog/pg_authid.h"
#include "catalog/pg_type.h"
#include "funcapi.h"
#include "miscadmin.h"
#include "pgstat.h"
+#include "storage/spin.h"
+#include "utils/acl.h"
#include "replication/walreceiver.h"
#include "storage/fd.h"
#include "storage/latch.h"
@@ -748,3 +751,98 @@ pg_promote(PG_FUNCTION_ARGS)
wait_seconds)));
PG_RETURN_BOOL(false);
}
+
+/*
+ * pg_stat_get_recovery - returns information about WAL recovery state
+ *
+ * Returns NULL when not in recovery or when the caller lacks
+ * pg_read_all_stats privileges; one row otherwise.
+ */
+Datum
+pg_stat_get_recovery(PG_FUNCTION_ARGS)
+{
+ TupleDesc tupdesc;
+ Datum *values;
+ bool *nulls;
+
+ /* Local copies of shared state */
+ bool promote_triggered;
+ XLogRecPtr last_replayed_read_lsn;
+ XLogRecPtr last_replayed_end_lsn;
+ TimeLineID last_replayed_tli;
+ XLogRecPtr replay_end_lsn;
+ TimeLineID replay_end_tli;
+ TimestampTz current_chunk_start_time;
+ RecoveryPauseState pause_state;
+
+ if (!RecoveryInProgress())
+ PG_RETURN_NULL();
+
+ if (!has_privs_of_role(GetUserId(), ROLE_PG_READ_ALL_STATS))
+ PG_RETURN_NULL();
+
+ /* Take a lock to ensure value consistency */
+ SpinLockAcquire(&XLogRecoveryCtl->info_lck);
+ promote_triggered = XLogRecoveryCtl->SharedPromoteIsTriggered;
+ last_replayed_read_lsn = XLogRecoveryCtl->lastReplayedReadRecPtr;
+ last_replayed_end_lsn = XLogRecoveryCtl->lastReplayedEndRecPtr;
+ last_replayed_tli = XLogRecoveryCtl->lastReplayedTLI;
+ replay_end_lsn = XLogRecoveryCtl->replayEndRecPtr;
+ replay_end_tli = XLogRecoveryCtl->replayEndTLI;
+ current_chunk_start_time = XLogRecoveryCtl->currentChunkStartTime;
+ pause_state = XLogRecoveryCtl->recoveryPauseState;
+ SpinLockRelease(&XLogRecoveryCtl->info_lck);
+
+ if (get_call_result_type(fcinfo, NULL, &tupdesc) != TYPEFUNC_COMPOSITE)
+ elog(ERROR, "return type must be a row type");
+
+ values = palloc0_array(Datum, tupdesc->natts);
+ nulls = palloc0_array(bool, tupdesc->natts);
+
+ values[0] = BoolGetDatum(promote_triggered);
+
+ if (XLogRecPtrIsValid(last_replayed_read_lsn))
+ values[1] = LSNGetDatum(last_replayed_read_lsn);
+ else
+ nulls[1] = true;
+
+ if (XLogRecPtrIsValid(last_replayed_end_lsn))
+ values[2] = LSNGetDatum(last_replayed_end_lsn);
+ else
+ nulls[2] = true;
+
+ if (XLogRecPtrIsValid(last_replayed_end_lsn))
+ values[3] = Int32GetDatum(last_replayed_tli);
+ else
+ nulls[3] = true;
+
+ if (XLogRecPtrIsValid(replay_end_lsn))
+ values[4] = LSNGetDatum(replay_end_lsn);
+ else
+ nulls[4] = true;
+
+ if (XLogRecPtrIsValid(replay_end_lsn))
+ values[5] = Int32GetDatum(replay_end_tli);
+ else
+ nulls[5] = true;
+
+ if (current_chunk_start_time != 0)
+ values[6] = TimestampTzGetDatum(current_chunk_start_time);
+ else
+ nulls[6] = true;
+
+ switch (pause_state)
+ {
+ case RECOVERY_NOT_PAUSED:
+ values[7] = CStringGetTextDatum("not paused");
+ break;
+ case RECOVERY_PAUSE_REQUESTED:
+ values[7] = CStringGetTextDatum("pause requested");
+ break;
+ case RECOVERY_PAUSED:
+ values[7] = CStringGetTextDatum("paused");
+ break;
+ }
+
+ PG_RETURN_DATUM(HeapTupleGetDatum(heap_form_tuple(tupdesc, values, nulls)));
+}
diff --git a/src/backend/catalog/system_views.sql b/src/backend/catalog/system_views.sql
index 1ea8f1faa9e..3aa77558bd6 100644
--- a/src/backend/catalog/system_views.sql
+++ b/src/backend/catalog/system_views.sql
@@ -1005,6 +1005,19 @@ CREATE VIEW pg_stat_wal_receiver AS
FROM pg_stat_get_wal_receiver() s
WHERE s.pid IS NOT NULL;
+CREATE VIEW pg_stat_recovery AS
+ SELECT
+ s.promote_triggered,
+ s.last_replayed_read_lsn,
+ s.last_replayed_end_lsn,
+ s.last_replayed_tli,
+ s.replay_end_lsn,
+ s.replay_end_tli,
+ s.current_chunk_start_time,
+ s.pause_state
+ FROM pg_stat_get_recovery() s
+ WHERE s.promote_triggered IS NOT NULL;
+
CREATE VIEW pg_stat_recovery_prefetch AS
SELECT
s.stats_reset,
diff --git a/src/include/catalog/pg_proc.dat b/src/include/catalog/pg_proc.dat
index dac40992cbc..514ec8dc5ad 100644
--- a/src/include/catalog/pg_proc.dat
+++ b/src/include/catalog/pg_proc.dat
@@ -5698,6 +5698,13 @@
proargmodes => '{o,o,o,o,o,o,o,o,o,o,o,o,o,o,o}',
proargnames => '{pid,status,receive_start_lsn,receive_start_tli,written_lsn,flushed_lsn,received_tli,last_msg_send_time,last_msg_receipt_time,latest_end_lsn,latest_end_time,slot_name,sender_host,sender_port,conninfo}',
prosrc => 'pg_stat_get_wal_receiver' },
+{ oid => '9949', descr => 'statistics: information about WAL recovery',
+ proname => 'pg_stat_get_recovery', proisstrict => 'f', provolatile => 's',
+ proparallel => 'r', prorettype => 'record', proargtypes => '',
+ proallargtypes => '{bool,pg_lsn,pg_lsn,int4,pg_lsn,int4,timestamptz,text}',
+ proargmodes => '{o,o,o,o,o,o,o,o}',
+ proargnames => '{promote_triggered,last_replayed_read_lsn,last_replayed_end_lsn,last_replayed_tli,replay_end_lsn,replay_end_tli,current_chunk_start_time,pause_state}',
+ prosrc => 'pg_stat_get_recovery' },
{ oid => '6169', descr => 'statistics: information about replication slot',
proname => 'pg_stat_get_replication_slot', provolatile => 's',
proparallel => 'r', prorettype => 'record', proargtypes => 'text',
diff --git a/src/test/regress/expected/rules.out b/src/test/regress/expected/rules.out
index 78a37d9fc8f..1150dc0ebf2 100644
--- a/src/test/regress/expected/rules.out
+++ b/src/test/regress/expected/rules.out
@@ -2127,6 +2127,16 @@ pg_stat_progress_vacuum| SELECT s.pid,
END AS started_by
FROM (pg_stat_get_progress_info('VACUUM'::text) s(pid, datid, relid, param1, param2, param3, param4, param5, param6, param7, param8, param9, param10, param11, param12, param13, param14, param15, param16, param17, param18, param19, param20)
LEFT JOIN pg_database d ON ((s.datid = d.oid)));
+pg_stat_recovery| SELECT promote_triggered,
+ last_replayed_read_lsn,
+ last_replayed_end_lsn,
+ last_replayed_tli,
+ replay_end_lsn,
+ replay_end_tli,
+ current_chunk_start_time,
+ pause_state
+ FROM pg_stat_get_recovery() s(promote_triggered, last_replayed_read_lsn, last_replayed_end_lsn, last_replayed_tli, replay_end_lsn, replay_end_tli, current_chunk_start_time, pause_state)
+ WHERE (s.promote_triggered IS NOT NULL);
pg_stat_recovery_prefetch| SELECT stats_reset,
prefetch,
hit,
diff --git a/src/test/regress/expected/sysviews.out b/src/test/regress/expected/sysviews.out
index 3dd63fd88ed..132b56a5864 100644
--- a/src/test/regress/expected/sysviews.out
+++ b/src/test/regress/expected/sysviews.out
@@ -143,6 +143,13 @@ select count(*) = 0 as ok from pg_stat_wal_receiver;
t
(1 row)
+-- We expect no recovery state in this test (running on primary)
+select count(*) = 0 as ok from pg_stat_recovery;
+ ok
+----
+ t
+(1 row)
+
-- This is to record the prevailing planner enable_foo settings during
-- a regression test run.
select name, setting from pg_settings where name like 'enable%';
diff --git a/src/test/regress/sql/sysviews.sql b/src/test/regress/sql/sysviews.sql
index 004f9a70e00..507e400ad4a 100644
--- a/src/test/regress/sql/sysviews.sql
+++ b/src/test/regress/sql/sysviews.sql
@@ -76,6 +76,9 @@ select count(*) = 1 as ok from pg_stat_wal;
-- We expect no walreceiver running in this test
select count(*) = 0 as ok from pg_stat_wal_receiver;
+-- We expect no recovery state in this test (running on primary)
+select count(*) = 0 as ok from pg_stat_recovery;
+
-- This is to record the prevailing planner enable_foo settings during
-- a regression test run.
select name, setting from pg_settings where name like 'enable%';
--
2.51.0