On Thu, Feb 4, 2021 at 6:16 PM Dilip Kumar <dilipbal...@gmail.com> wrote:
>
> On Thu, Feb 4, 2021 at 4:58 PM Bharath Rupireddy
> <bharath.rupireddyforpostg...@gmail.com> wrote:
> >
> > On Thu, Feb 4, 2021 at 10:28 AM Dilip Kumar <dilipbal...@gmail.com> wrote:
> > > Please find an updated patch which addresses these comments.
> >
> > Thanks for the patch. I tested the new function
> > pg_get_wal_replay_pause_state:
> >
> > postgres=# select pg_get_wal_replay_pause_state();
> > pg_get_wal_replay_pause_state
> > -------------------------------
> > not paused
> > postgres=# select pg_wal_replay_pause();
> > pg_wal_replay_pause
> > ---------------------
> >
> > (1 row)
> >
> > I can also see the "pause requested" state after I put a gdb
> > breakpoint in WaitForWALToBecomeAvailable in the standby startup
> > process .
> >
> > postgres=# select pg_get_wal_replay_pause_state();
> > pg_get_wal_replay_pause_state
> > -------------------------------
> > pause requested
> > (1 row)
> >
> > postgres=# select pg_get_wal_replay_pause_state();
> > pg_get_wal_replay_pause_state
> > -------------------------------
> > paused
> > (1 row)
> >
> > Mostly, the v10 patch looks good to me, except below minor comments:
> >
> > 1) A typo in commit message - "just check" --> "just checks"
> >
> > 2) How about
> > + Returns recovery pause state. The return values are <literal>not
> > paused
> > instead of
> > + Returns recovery pause state, the return values are <literal>not
> > paused
> >
> > 3) I think it is 'get wal replay pause state', instead of { oid =>
> > '1137', descr => 'get wal replay is pause state',
> >
> > 4) can we just do this
> > /*
> > * If recovery pause is requested then set it paused. While we are
> > in
> > * the loop, user might resume and pause again so set this every
> > time.
> > */
> > if (((volatile XLogCtlData *) XLogCtl)->recoveryPauseState ==
> > RECOVERY_PAUSE_REQUESTED)
> > SetRecoveryPause(RECOVERY_PAUSED);
> > instead of
> > /*
> > * If recovery pause is requested then set it paused. While we are
> > in
> > * the loop, user might resume and pause again so set this every
> > time.
> > */
> > SpinLockAcquire(&XLogCtl->info_lck);
> > if (XLogCtl->recoveryPauseState == RECOVERY_PAUSE_REQUESTED)
> > XLogCtl->recoveryPauseState = RECOVERY_PAUSED;
> > SpinLockRelease(&XLogCtl->info_lck);
> >
> > I think it's okay, since we take a spinlock anyways in
> > GetRecoveryPauseState(). See the below comment and also a relevant
> > commit 6ba4ecbf477e0b25dd7bde1b0c4e07fc2da19348 on why it's not
> > necessary taking spinlock always:
> > /*
> > * Pause WAL replay, if requested by a hot-standby session
> > via
> > * SetRecoveryPause().
> > *
> > * Note that we intentionally don't take the info_lck
> > spinlock
> > * here. We might therefore read a slightly stale value of
> > * the recoveryPause flag, but it can't be very stale (no
> > * worse than the last spinlock we did acquire). Since a
> > * pause request is a pretty asynchronous thing anyway,
> > * possibly responding to it one WAL record later than we
> > * otherwise would is a minor issue, so it doesn't seem
> > worth
> > * adding another spinlock cycle to prevent that.
> > */
>
> How can we do that this is not a 1 byte flag this is enum so I don't
> think we can read any atomic state without a spin lock here.
I have fixed the other comments and the updated patch is attached.
--
Regards,
Dilip Kumar
EnterpriseDB: http://www.enterprisedb.com
From 0d7059a8809cbd28e7a4668c6b99e0c297c564f3 Mon Sep 17 00:00:00 2001
From: Dilip Kumar <dilipkumar@localhost.localdomain>
Date: Wed, 27 Jan 2021 16:46:04 +0530
Subject: [PATCH v11] Provide a new interface to get the recovery pause status
Currently, pg_is_wal_replay_paused, just checks whether the recovery
pause is requested or not but it doesn't actually tell whether the
recovery is actually paused or not. So basically there is no way for
the user to know the actual status of the pause request. This patch
provides a new interface pg_get_wal_replay_pause_state that will
return the actual status of the recovery pause i.e.'not paused' if
pause is not requested, 'pause requested' if pause is requested but
recovery is not yet paused and 'paused' if recovery is actually paused.
---
doc/src/sgml/func.sgml | 32 ++++++++++++++---
src/backend/access/transam/xlog.c | 64 +++++++++++++++++++++++++---------
src/backend/access/transam/xlogfuncs.c | 50 +++++++++++++++++++++++---
src/include/access/xlog.h | 12 +++++--
src/include/catalog/pg_proc.dat | 4 +++
5 files changed, 134 insertions(+), 28 deletions(-)
diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index f30eaa3..eb131a6 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -25251,7 +25251,24 @@ postgres=# SELECT * FROM pg_walfile_name_offset(pg_stop_backup());
<returnvalue>boolean</returnvalue>
</para>
<para>
- Returns true if recovery is paused.
+ Returns true if recovery pause is requested.
+ </para></entry>
+ </row>
+
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>pg_get_wal_replay_pause_state</primary>
+ </indexterm>
+ <function>pg_get_wal_replay_pause_state</function> ()
+ <returnvalue>text</returnvalue>
+ </para>
+ <para>
+ Returns recovery pause state. The return values are <literal>
+ not paused</literal> if pause is not requested, <literal>
+ pause requested</literal> if pause is requested but recovery is
+ not yet paused and, <literal>paused</literal> if the recovery is
+ actually paused.
</para></entry>
</row>
@@ -25290,10 +25307,15 @@ postgres=# SELECT * FROM pg_walfile_name_offset(pg_stop_backup());
<returnvalue>void</returnvalue>
</para>
<para>
- Pauses recovery. While recovery is paused, no further database
- changes are applied. If hot standby is active, all new queries will
- see the same consistent snapshot of the database, and no further query
- conflicts will be generated until recovery is resumed.
+ Request to pause recovery. A request doesn't mean that recovery stops
+ right away. If you want a guarantee that recovery is actually paused,
+ you need to check for the recovery pause state returned by
+ <function>pg_get_wal_replay_pause_state()</function>. Note that
+ <function>pg_is_wal_repay_paused()</function> returns whether a request
+ is made. While recovery is paused, no further database changes are applied.
+ If hot standby is active, all new queries will see the same consistent
+ snapshot of the database, and no further query conflicts will be generated
+ until recovery is resumed.
</para>
<para>
This function is restricted to superusers by default, but other users
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index f03bd47..fb16293 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -721,8 +721,8 @@ typedef struct XLogCtlData
* only relevant for replication or archive recovery
*/
TimestampTz currentChunkStartTime;
- /* Are we requested to pause recovery? */
- bool recoveryPause;
+ /* Recovery pause state */
+ RecoveryPauseState recoveryPauseState;
/*
* lastFpwDisableRecPtr points to the start of the last replayed
@@ -6019,7 +6019,7 @@ recoveryStopsAfter(XLogReaderState *record)
}
/*
- * Wait until shared recoveryPause flag is cleared.
+ * Wait until shared recoveryPauseState is set to RECOVERY_NOT_PAUSED.
*
* endOfRecovery is true if the recovery target is reached and
* the paused state starts at the end of recovery because of
@@ -6049,34 +6049,55 @@ recoveryPausesHere(bool endOfRecovery)
(errmsg("recovery has paused"),
errhint("Execute pg_wal_replay_resume() to continue.")));
- while (RecoveryIsPaused())
+ /* loop until recoveryPauseState is set to RECOVERY_NOT_PAUSED */
+ while (GetRecoveryPauseState() != RECOVERY_NOT_PAUSED)
{
HandleStartupProcInterrupts();
+
if (CheckForStandbyTrigger())
return;
pgstat_report_wait_start(WAIT_EVENT_RECOVERY_PAUSE);
+
+ /*
+ * If recovery pause is requested then set it paused. While we are in
+ * the loop, user might resume and pause again so set this every time.
+ */
+ SpinLockAcquire(&XLogCtl->info_lck);
+ if (XLogCtl->recoveryPauseState == RECOVERY_PAUSE_REQUESTED)
+ XLogCtl->recoveryPauseState = RECOVERY_PAUSED;
+ SpinLockRelease(&XLogCtl->info_lck);
+
pg_usleep(1000000L); /* 1000 ms */
pgstat_report_wait_end();
}
}
-bool
-RecoveryIsPaused(void)
+/*
+ * Get the current state of the recovery pause request.
+ */
+RecoveryPauseState
+GetRecoveryPauseState(void)
{
- bool recoveryPause;
+ RecoveryPauseState state;
SpinLockAcquire(&XLogCtl->info_lck);
- recoveryPause = XLogCtl->recoveryPause;
+ state = XLogCtl->recoveryPauseState;
SpinLockRelease(&XLogCtl->info_lck);
- return recoveryPause;
+ return state;
}
+/*
+ * Set the recovery pause state.
+ */
void
-SetRecoveryPause(bool recoveryPause)
+SetRecoveryPause(RecoveryPauseState state)
{
+ if (state < RECOVERY_NOT_PAUSED || state > RECOVERY_PAUSED)
+ elog(ERROR, "invalid recovery pause state %d", state);
+
SpinLockAcquire(&XLogCtl->info_lck);
- XLogCtl->recoveryPause = recoveryPause;
+ XLogCtl->recoveryPauseState = state;
SpinLockRelease(&XLogCtl->info_lck);
}
@@ -6270,14 +6291,14 @@ RecoveryRequiresIntParameter(const char *param_name, int currValue, int minValue
currValue,
minValue)));
- SetRecoveryPause(true);
+ SetRecoveryPause(RECOVERY_PAUSED);
ereport(LOG,
(errmsg("recovery has paused"),
errdetail("If recovery is unpaused, the server will shut down."),
errhint("You can then restart the server after making the necessary configuration changes.")));
- while (RecoveryIsPaused())
+ while (GetRecoveryPauseState() != RECOVERY_NOT_PAUSED)
{
HandleStartupProcInterrupts();
@@ -7194,7 +7215,7 @@ StartupXLOG(void)
XLogCtl->lastReplayedTLI = XLogCtl->replayEndTLI;
XLogCtl->recoveryLastXTime = 0;
XLogCtl->currentChunkStartTime = 0;
- XLogCtl->recoveryPause = false;
+ XLogCtl->recoveryPauseState = RECOVERY_NOT_PAUSED;
SpinLockRelease(&XLogCtl->info_lck);
/* Also ensure XLogReceiptTime has a sane value */
@@ -7298,7 +7319,8 @@ StartupXLOG(void)
* otherwise would is a minor issue, so it doesn't seem worth
* adding another spinlock cycle to prevent that.
*/
- if (((volatile XLogCtlData *) XLogCtl)->recoveryPause)
+ if (((volatile XLogCtlData *) XLogCtl)->recoveryPauseState ==
+ RECOVERY_PAUSE_REQUESTED)
recoveryPausesHere(false);
/*
@@ -7323,7 +7345,8 @@ StartupXLOG(void)
* here otherwise pausing during the delay-wait wouldn't
* work.
*/
- if (((volatile XLogCtlData *) XLogCtl)->recoveryPause)
+ if (((volatile XLogCtlData *) XLogCtl)->recoveryPauseState ==
+ RECOVERY_PAUSE_REQUESTED)
recoveryPausesHere(false);
}
@@ -7497,7 +7520,7 @@ StartupXLOG(void)
proc_exit(3);
case RECOVERY_TARGET_ACTION_PAUSE:
- SetRecoveryPause(true);
+ SetRecoveryPause(RECOVERY_PAUSE_REQUESTED);
recoveryPausesHere(true);
/* drop into promote */
@@ -12625,6 +12648,13 @@ WaitForWALToBecomeAvailable(XLogRecPtr RecPtr, bool randAccess,
elog(ERROR, "unexpected WAL source %d", currentSource);
}
+ /* test for recovery pause, if user has requested the pause */
+ if (((volatile XLogCtlData *) XLogCtl)->recoveryPauseState ==
+ RECOVERY_PAUSE_REQUESTED)
+ recoveryPausesHere(false);
+
+ now = GetCurrentTimestamp();
+
/*
* This possibly-long loop needs to handle interrupts of startup
* process.
diff --git a/src/backend/access/transam/xlogfuncs.c b/src/backend/access/transam/xlogfuncs.c
index 5e1aab3..349ad0c 100644
--- a/src/backend/access/transam/xlogfuncs.c
+++ b/src/backend/access/transam/xlogfuncs.c
@@ -517,7 +517,7 @@ pg_walfile_name(PG_FUNCTION_ARGS)
}
/*
- * pg_wal_replay_pause - pause recovery now
+ * pg_wal_replay_pause - Request to pause recovery
*
* Permission checking for this function is managed through the normal
* GRANT system.
@@ -538,7 +538,10 @@ pg_wal_replay_pause(PG_FUNCTION_ARGS)
errhint("%s cannot be executed after promotion is triggered.",
"pg_wal_replay_pause()")));
- SetRecoveryPause(true);
+ SetRecoveryPause(RECOVERY_PAUSE_REQUESTED);
+
+ /* wake up the recovery process so that it can process the pause request */
+ WakeupRecovery();
PG_RETURN_VOID();
}
@@ -565,7 +568,7 @@ pg_wal_replay_resume(PG_FUNCTION_ARGS)
errhint("%s cannot be executed after promotion is triggered.",
"pg_wal_replay_resume()")));
- SetRecoveryPause(false);
+ SetRecoveryPause(RECOVERY_NOT_PAUSED);
PG_RETURN_VOID();
}
@@ -582,7 +585,46 @@ pg_is_wal_replay_paused(PG_FUNCTION_ARGS)
errmsg("recovery is not in progress"),
errhint("Recovery control functions can only be executed during recovery.")));
- PG_RETURN_BOOL(RecoveryIsPaused());
+ PG_RETURN_BOOL(GetRecoveryPauseState() != RECOVERY_NOT_PAUSED);
+}
+
+/*
+ * pg_get_wal_replay_pause_state - Returns the recovery pause state.
+ *
+ * Returned values:
+ *
+ * 'not paused' - if pause is not requested
+ * 'pause requested' - if pause is requested but recovery is not yet paused
+ * 'paused' - if recovery is paused
+ */
+Datum
+pg_get_wal_replay_pause_state(PG_FUNCTION_ARGS)
+{
+ char *state;
+
+ 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.")));
+
+ /* get the recovery pause state */
+ switch(GetRecoveryPauseState())
+ {
+ case RECOVERY_NOT_PAUSED:
+ state = "not paused";
+ break;
+ case RECOVERY_PAUSE_REQUESTED:
+ state = "pause requested";
+ break;
+ case RECOVERY_PAUSED:
+ state = "paused";
+ break;
+ default:
+ elog(ERROR, "invalid recovery pause state");
+ }
+
+ PG_RETURN_TEXT_P(cstring_to_text(state));
}
/*
diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h
index 75ec107..7533c48 100644
--- a/src/include/access/xlog.h
+++ b/src/include/access/xlog.h
@@ -174,6 +174,14 @@ typedef enum RecoveryState
RECOVERY_STATE_DONE /* currently in production */
} RecoveryState;
+/* Recovery pause states */
+typedef enum RecoveryPauseState
+{
+ RECOVERY_NOT_PAUSED = 0, /* pause not requested */
+ RECOVERY_PAUSE_REQUESTED = 1, /* pause requested, but yet paused */
+ RECOVERY_PAUSED = 2 /* recovery is paused */
+} RecoveryPauseState;
+
extern PGDLLIMPORT int wal_level;
/* Is WAL archiving enabled (always or only while server is running normally)? */
@@ -310,8 +318,8 @@ extern void GetXLogReceiptTime(TimestampTz *rtime, bool *fromStream);
extern XLogRecPtr GetXLogReplayRecPtr(TimeLineID *replayTLI);
extern XLogRecPtr GetXLogInsertRecPtr(void);
extern XLogRecPtr GetXLogWriteRecPtr(void);
-extern bool RecoveryIsPaused(void);
-extern void SetRecoveryPause(bool recoveryPause);
+extern RecoveryPauseState GetRecoveryPauseState(void);
+extern void SetRecoveryPause(RecoveryPauseState state);
extern TimestampTz GetLatestXTime(void);
extern TimestampTz GetCurrentChunkReplayStartTime(void);
diff --git a/src/include/catalog/pg_proc.dat b/src/include/catalog/pg_proc.dat
index b5f52d4..80d3c7a 100644
--- a/src/include/catalog/pg_proc.dat
+++ b/src/include/catalog/pg_proc.dat
@@ -6222,6 +6222,10 @@
proname => 'pg_is_wal_replay_paused', provolatile => 'v',
prorettype => 'bool', proargtypes => '',
prosrc => 'pg_is_wal_replay_paused' },
+{ oid => '1137', descr => 'get wal replay pause state',
+ proname => 'pg_get_wal_replay_pause_state', provolatile => 'v',
+ prorettype => 'text', proargtypes => '',
+ prosrc => 'pg_get_wal_replay_pause_state' },
{ oid => '2621', descr => 'reload configuration files',
proname => 'pg_reload_conf', provolatile => 'v', prorettype => 'bool',
--
1.8.3.1
