On Sat, Oct 20, 2018 at 01:48:56PM +0900, Michael Paquier wrote: > On Sat, Oct 20, 2018 at 06:24:28AM +0200, Laurenz Albe wrote: >> Here is another version, with a fix in pg_proc.dat, an improved comment >> and "wait_seconds" exercised in the regression test. > > Thanks for the new version. This looks pretty good to me. I'll see if > I can review it once and then commit. > >> - WAIT_EVENT_SYNC_REP >> + WAIT_EVENT_SYNC_REP, >> + WAIT_EVENT_PROMOTE >> } WaitEventIPC; > > Those are kept in alphabetical order. Individual wait events are also > documented with a description.
Regarding the documentation, wouldn't it be more adapted to list the new function under the section "Recovery Control Functions"? Not only does the new function signal the postmaster, but it also creates the promotion trigger file. Anyway, I have been looking in depth at the patch, and I finish with the attached. Here are some notes: - pg_ctl returns an error if it cannot create the promote trigger file and if it doesn't close it. pg_promote should do the same. - WL_TIMEOUT could have been reached, leading to no further retries (remove for example the part related to the trigger file creation and try to trigger the promotion, the wait time is incorrect). - Documentation has been reworked. - The new wait event is documented. - a WARNING is returned if the signal to the postmaster is not sent, which is something I think makes most sense as we do the same for other signals. - position including unistd.h was not correct in xlogfuncs.c. - Timeouts for the tests are made a tad longer. Some buildfarm machines don't like a promotion wait of 10s. - a catalog version bump is included, just a personal reminder.. - Indentatio has been run. Laurenz, what do you think? -- Michael
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 5193df3366..96d45419e5 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -19202,6 +19202,9 @@ postgres=# SELECT * FROM pg_walfile_name_offset(pg_stop_backup());
<indexterm>
<primary>pg_is_wal_replay_paused</primary>
</indexterm>
+ <indexterm>
+ <primary>pg_promote</primary>
+ </indexterm>
<indexterm>
<primary>pg_wal_replay_pause</primary>
</indexterm>
@@ -19232,6 +19235,22 @@ postgres=# SELECT * FROM pg_walfile_name_offset(pg_stop_backup());
<entry>True if recovery is paused.
</entry>
</row>
+ <row>
+ <entry>
+ <literal><function>pg_promote(<parameter>wait</parameter> <type>boolean</type> DEFAULT true, <parameter>wait_seconds</parameter> <type>integer</type> DEFAULT 60)</function></literal>
+ </entry>
+ <entry><type>boolean</type></entry>
+ <entry>
+ Promotes a physical standby server. Returns <literal>true</literal>
+ if promotion is successful and <literal>false</literal> otherwise.
+ With <parameter>wait</parameter> set to <literal>true</literal>, the
+ default, the function waits until promotion is completed or
+ <parameter>wait_seconds</parameter> seconds have passed, otherwise the
+ function returns immediately after sending the promotion signal to the
+ postmaster. This function is restricted to superusers by default, but
+ other users can be granted EXECUTE to run the function.
+ </entry>
+ </row>
<row>
<entry>
<literal><function>pg_wal_replay_pause()</function></literal>
diff --git a/doc/src/sgml/high-availability.sgml b/doc/src/sgml/high-availability.sgml
index ebcb3daaed..faf8e71854 100644
--- a/doc/src/sgml/high-availability.sgml
+++ b/doc/src/sgml/high-availability.sgml
@@ -1471,14 +1471,17 @@ synchronous_standby_names = 'ANY 2 (s1, s2, s3)'
</para>
<para>
- To trigger failover of a log-shipping standby server,
- run <command>pg_ctl promote</command> or create a trigger
- file with the file name and path specified by the <varname>trigger_file</varname>
- setting in <filename>recovery.conf</filename>. If you're planning to use
- <command>pg_ctl promote</command> to fail over, <varname>trigger_file</varname> is
- not required. If you're setting up the reporting servers that are
- only used to offload read-only queries from the primary, not for high
- availability purposes, you don't need to promote it.
+ To trigger failover of a log-shipping standby server, run
+ <command>pg_ctl promote</command>, call <function>pg_promote</function>,
+ or create a trigger file with the file name and path specified by the
+ <varname>trigger_file</varname> setting in
+ <filename>recovery.conf</filename>. If you're planning to use
+ <command>pg_ctl promote</command> or to call
+ <function>pg_promote</function> to fail over,
+ <varname>trigger_file</varname> is not required. If you're
+ setting up the reporting servers that are only used to offload read-only
+ queries from the primary, not for high availability purposes, you don't
+ need to promote it.
</para>
</sect1>
diff --git a/doc/src/sgml/monitoring.sgml b/doc/src/sgml/monitoring.sgml
index 0484cfa77a..073f19542a 100644
--- a/doc/src/sgml/monitoring.sgml
+++ b/doc/src/sgml/monitoring.sgml
@@ -1268,7 +1268,7 @@ postgres 27093 0.0 0.0 30096 2752 ? Ss 11:34 0:00 postgres: ser
<entry>Waiting in an extension.</entry>
</row>
<row>
- <entry morerows="33"><literal>IPC</literal></entry>
+ <entry morerows="34"><literal>IPC</literal></entry>
<entry><literal>BgWorkerShutdown</literal></entry>
<entry>Waiting for background worker to shut down.</entry>
</row>
@@ -1384,6 +1384,10 @@ postgres 27093 0.0 0.0 30096 2752 ? Ss 11:34 0:00 postgres: ser
<entry><literal>ProcArrayGroupUpdate</literal></entry>
<entry>Waiting for group leader to clear transaction id at transaction end.</entry>
</row>
+ <row>
+ <entry><literal>Promote</literal></entry>
+ <entry>Waiting for standby promotion.</entry>
+ </row>
<row>
<entry><literal>ClogGroupUpdate</literal></entry>
<entry>Waiting for group leader to update transaction status at transaction end.</entry>
diff --git a/doc/src/sgml/recovery-config.sgml b/doc/src/sgml/recovery-config.sgml
index 92825fdf19..a2bdffda94 100644
--- a/doc/src/sgml/recovery-config.sgml
+++ b/doc/src/sgml/recovery-config.sgml
@@ -439,7 +439,8 @@ restore_command = 'copy "C:\\server\\archivedir\\%f" "%p"' # Windows
<para>
Specifies a trigger file whose presence ends recovery in the
standby. Even if this value is not set, you can still promote
- the standby using <command>pg_ctl promote</command>.
+ the standby using <command>pg_ctl promote</command> or calling
+ <function>pg_promote</function>.
This setting has no effect if <varname>standby_mode</varname> is <literal>off</literal>.
</para>
</listitem>
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 7375a78ffc..62fc418893 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -78,12 +78,6 @@
extern uint32 bootstrap_data_checksum_version;
-/* File path names (all relative to $PGDATA) */
-#define RECOVERY_COMMAND_FILE "recovery.conf"
-#define RECOVERY_COMMAND_DONE "recovery.done"
-#define PROMOTE_SIGNAL_FILE "promote"
-#define FALLBACK_PROMOTE_SIGNAL_FILE "fallback_promote"
-
/* User-settable parameters */
int max_wal_size_mb = 1024; /* 1 GB */
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 9731742978..6279b17caa 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -16,6 +16,8 @@
*/
#include "postgres.h"
+#include <unistd.h>
+
#include "access/htup_details.h"
#include "access/xlog.h"
#include "access/xlog_internal.h"
@@ -23,6 +25,7 @@
#include "catalog/pg_type.h"
#include "funcapi.h"
#include "miscadmin.h"
+#include "pgstat.h"
#include "replication/walreceiver.h"
#include "storage/smgr.h"
#include "utils/builtins.h"
@@ -697,3 +700,77 @@ pg_backup_start_time(PG_FUNCTION_ARGS)
PG_RETURN_DATUM(xtime);
}
+
+/*
+ * Promotes a standby server.
+ *
+ * A result of "true" means that promotion has been completed if "wait is
+ * "true", or initiated if "wait" is false.
+ */
+Datum
+pg_promote(PG_FUNCTION_ARGS)
+{
+ bool wait = PG_GETARG_BOOL(0);
+ int wait_seconds = PG_GETARG_INT32(1);
+ FILE *promote_file;
+ int i;
+
+ if (!RecoveryInProgress())
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("recovery is not in progress"),
+ errhint("Recovery control functions can only be executed during recovery.")));
+
+ if (wait_seconds <= 0)
+ ereport(ERROR,
+ (errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE),
+ errmsg("\"wait_seconds\" cannot be negative or equal zero")));
+
+ /* create the promote signal file */
+ promote_file = AllocateFile(PROMOTE_SIGNAL_FILE, "w");
+ if (!promote_file)
+ ereport(ERROR,
+ (errcode_for_file_access(),
+ errmsg("could not create file \"%s\": %m",
+ PROMOTE_SIGNAL_FILE)));
+
+ if (FreeFile(promote_file))
+ ereport(ERROR,
+ (errcode_for_file_access(),
+ errmsg("could not write file \"%s\": %m",
+ PROMOTE_SIGNAL_FILE)));
+
+ /* signal the postmaster */
+ if (kill(PostmasterPid, SIGUSR1) != 0)
+ {
+ ereport(WARNING,
+ (errmsg("failed to send signal to postmaster: %m")));
+ (void) unlink(PROMOTE_SIGNAL_FILE);
+ PG_RETURN_BOOL(false);
+ }
+
+ /* return immediately if waiting was not requested */
+ if (!wait)
+ PG_RETURN_BOOL(true);
+
+ /* wait for the amount of time wanted until promotion */
+#define WAITS_PER_SECOND 10
+ for (i = 0; i < WAITS_PER_SECOND * wait_seconds; ++i)
+ {
+ ResetLatch(MyLatch);
+
+ if (!RecoveryInProgress())
+ PG_RETURN_BOOL(true);
+
+ CHECK_FOR_INTERRUPTS();
+
+ WaitLatch(MyLatch,
+ WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH,
+ 1000L / WAITS_PER_SECOND,
+ WAIT_EVENT_PROMOTE);
+ }
+
+ ereport(WARNING,
+ (errmsg("server did not promote within %d seconds", wait_seconds)));
+ PG_RETURN_BOOL(false);
+}
diff --git a/src/backend/catalog/system_views.sql b/src/backend/catalog/system_views.sql
index a03b005f73..2857d80984 100644
--- a/src/backend/catalog/system_views.sql
+++ b/src/backend/catalog/system_views.sql
@@ -1119,6 +1119,13 @@ LANGUAGE INTERNAL
STRICT IMMUTABLE PARALLEL SAFE
AS 'jsonb_insert';
+CREATE OR REPLACE FUNCTION
+ pg_promote(wait boolean DEFAULT true, wait_seconds integer DEFAULT 60)
+RETURNS boolean
+LANGUAGE INTERNAL
+STRICT VOLATILE
+AS 'pg_promote';
+
--
-- The default permissions for functions mean that anyone can execute them.
-- A number of functions shouldn't be executable by just anyone, but rather
@@ -1138,6 +1145,7 @@ REVOKE EXECUTE ON FUNCTION pg_rotate_logfile() FROM public;
REVOKE EXECUTE ON FUNCTION pg_reload_conf() FROM public;
REVOKE EXECUTE ON FUNCTION pg_current_logfile() FROM public;
REVOKE EXECUTE ON FUNCTION pg_current_logfile(text) FROM public;
+REVOKE EXECUTE ON FUNCTION pg_promote(boolean, integer) FROM public;
REVOKE EXECUTE ON FUNCTION pg_stat_reset() FROM public;
REVOKE EXECUTE ON FUNCTION pg_stat_reset_shared(text) FROM public;
diff --git a/src/backend/postmaster/pgstat.c b/src/backend/postmaster/pgstat.c
index 8de603d193..d02d6a647d 100644
--- a/src/backend/postmaster/pgstat.c
+++ b/src/backend/postmaster/pgstat.c
@@ -3660,6 +3660,9 @@ pgstat_get_wait_ipc(WaitEventIPC w)
case WAIT_EVENT_PROCARRAY_GROUP_UPDATE:
event_name = "ProcArrayGroupUpdate";
break;
+ case WAIT_EVENT_PROMOTE:
+ event_name = "Promote";
+ break;
case WAIT_EVENT_CLOG_GROUP_UPDATE:
event_name = "ClogGroupUpdate";
break;
diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h
index 421ba6d775..e01d12eb7c 100644
--- a/src/include/access/xlog.h
+++ b/src/include/access/xlog.h
@@ -319,10 +319,16 @@ extern void do_pg_abort_backup(void);
extern SessionBackupState get_backup_status(void);
/* File path names (all relative to $PGDATA) */
+#define RECOVERY_COMMAND_FILE "recovery.conf"
+#define RECOVERY_COMMAND_DONE "recovery.done"
#define BACKUP_LABEL_FILE "backup_label"
#define BACKUP_LABEL_OLD "backup_label.old"
#define TABLESPACE_MAP "tablespace_map"
#define TABLESPACE_MAP_OLD "tablespace_map.old"
+/* files to signal promotion to primary */
+#define PROMOTE_SIGNAL_FILE "promote"
+#define FALLBACK_PROMOTE_SIGNAL_FILE "fallback_promote"
+
#endif /* XLOG_H */
diff --git a/src/include/catalog/catversion.h b/src/include/catalog/catversion.h
index e04eabf683..7c2d2aa075 100644
--- a/src/include/catalog/catversion.h
+++ b/src/include/catalog/catversion.h
@@ -53,6 +53,6 @@
*/
/* yyyymmddN */
-#define CATALOG_VERSION_NO 201810111
+#define CATALOG_VERSION_NO 201810221
#endif
diff --git a/src/include/catalog/pg_proc.dat b/src/include/catalog/pg_proc.dat
index cff58ed2d8..4d7fe1b383 100644
--- a/src/include/catalog/pg_proc.dat
+++ b/src/include/catalog/pg_proc.dat
@@ -5824,6 +5824,10 @@
proname => 'pg_backup_start_time', provolatile => 's',
prorettype => 'timestamptz', proargtypes => '',
prosrc => 'pg_backup_start_time' },
+{ oid => '3436', descr => 'promote standby server',
+ proname => 'pg_promote', provolatile => 'v',
+ prorettype => 'bool', proargtypes => 'bool int4', proargnames => '{wait,wait_seconds}',
+ prosrc => 'pg_promote' },
{ oid => '2848', descr => 'switch to new wal file',
proname => 'pg_switch_wal', provolatile => 'v', prorettype => 'pg_lsn',
proargtypes => '', prosrc => 'pg_switch_wal' },
diff --git a/src/include/pgstat.h b/src/include/pgstat.h
index d59c24ae23..e0ec3f9086 100644
--- a/src/include/pgstat.h
+++ b/src/include/pgstat.h
@@ -828,6 +828,7 @@ typedef enum
WAIT_EVENT_PARALLEL_BITMAP_SCAN,
WAIT_EVENT_PARALLEL_CREATE_INDEX_SCAN,
WAIT_EVENT_PROCARRAY_GROUP_UPDATE,
+ WAIT_EVENT_PROMOTE,
WAIT_EVENT_CLOG_GROUP_UPDATE,
WAIT_EVENT_REPLICATION_ORIGIN_DROP,
WAIT_EVENT_REPLICATION_SLOT_DROP,
diff --git a/src/test/recovery/t/004_timeline_switch.pl b/src/test/recovery/t/004_timeline_switch.pl
index 34ee335129..e774ec26f0 100644
--- a/src/test/recovery/t/004_timeline_switch.pl
+++ b/src/test/recovery/t/004_timeline_switch.pl
@@ -37,9 +37,12 @@ $node_master->safe_psql('postgres',
$node_master->wait_for_catchup($node_standby_1, 'replay',
$node_master->lsn('write'));
-# Stop and remove master, and promote standby 1, switching it to a new timeline
+# Stop and remove master
$node_master->teardown_node;
-$node_standby_1->promote;
+
+# promote standby 1 using "pg_promote", switching it to a new timeline
+$node_standby_1->safe_psql('postgres',
+ "SELECT pg_promote(wait_seconds => 300)");
# Switch standby 2 to replay from standby 1
rmtree($node_standby_2->data_dir . '/recovery.conf');
diff --git a/src/test/recovery/t/009_twophase.pl b/src/test/recovery/t/009_twophase.pl
index 9ea3bd65fc..f5ebbd995e 100644
--- a/src/test/recovery/t/009_twophase.pl
+++ b/src/test/recovery/t/009_twophase.pl
@@ -214,7 +214,11 @@ $cur_master->psql(
INSERT INTO t_009_tbl VALUES (22, 'issued to ${cur_master_name}');
PREPARE TRANSACTION 'xact_009_10';");
$cur_master->teardown_node;
-$cur_standby->promote;
+
+# promote standby using "pg_promote" and wait until it is promoted
+$cur_standby->safe_psql('postgres', 'SELECT pg_promote(false)');
+$cur_standby->poll_query_until('postgres', "SELECT NOT pg_is_in_recovery()")
+ or die "standby never exited recovery";
# change roles
note "Now paris is master and london is standby";
signature.asc
Description: PGP signature
