#include "parser/parse_node.h"
#include "storage/block.h"
#include "storage/relfilelocator.h"
#include "storage/smgr.h"
#include "storage/sync.h"

Include dependency graph for bgwriter.h:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Functions
pg_noreturn void	BackgroundWriterMain (const void *startup_data, size_t startup_data_len)

pg_noreturn void	CheckpointerMain (const void *startup_data, size_t startup_data_len)

void	ExecCheckpoint (ParseState pstate, CheckPointStmt stmt)

void	RequestCheckpoint (int flags)

void	CheckpointWriteDelay (int flags, double progress)

bool	ForwardSyncRequest (const FileTag *ftag, SyncRequestType type)

void	AbsorbSyncRequests (void)

bool	FirstCallSinceLastCheckpoint (void)

Variables
PGDLLIMPORT int	BgWriterDelay

PGDLLIMPORT int	CheckPointTimeout

PGDLLIMPORT int	CheckPointWarning

PGDLLIMPORT double	CheckPointCompletionTarget

Function Documentation

◆ AbsorbSyncRequests()

void AbsorbSyncRequests ( void )

extern

Definition at line 1424 of file checkpointer.c.

{
    CheckpointerRequest *requests = NULL;
    CheckpointerRequest *request;
    int         n,
                i;
    bool        loop;
 
    if (!AmCheckpointerProcess())
        return;
 
    do
    {
        LWLockAcquire(CheckpointerCommLock, LW_EXCLUSIVE);
 
        /*---
         * We try to avoid holding the lock for a long time by:
         * 1. Copying the request array and processing the requests after
         *    releasing the lock;
         * 2. Processing not the whole queue, but only batches of
         *    CKPT_REQ_BATCH_SIZE at once.
         *
         * Once we have cleared the requests from shared memory, we must
         * PANIC if we then fail to absorb them (e.g., because our hashtable
         * runs out of memory).  This is because the system cannot run safely
         * if we are unable to fsync what we have been told to fsync.
         * Fortunately, the hashtable is so small that the problem is quite
         * unlikely to arise in practice.
         *
         * Note: The maximum possible size of a ring buffer is
         * MAX_CHECKPOINT_REQUESTS entries, which fit into a maximum palloc
         * allocation size of 1Gb.  Our maximum batch size,
         * CKPT_REQ_BATCH_SIZE, is even smaller.
         */
        n = Min(CheckpointerShmem->num_requests, CKPT_REQ_BATCH_SIZE);
        if (n > 0)
        {
            if (!requests)
                requests = (CheckpointerRequest *) palloc(n * sizeof(CheckpointerRequest));
 
            for (i = 0; i < n; i++)
            {
                requests[i] = CheckpointerShmem->requests[CheckpointerShmem->head];
                CheckpointerShmem->head = (CheckpointerShmem->head + 1) % CheckpointerShmem->max_requests;
            }
 
            CheckpointerShmem->num_requests -= n;
 
        }
 
        START_CRIT_SECTION();
 
        /* Are there any requests in the queue? If so, keep going. */
        loop = CheckpointerShmem->num_requests != 0;
 
        LWLockRelease(CheckpointerCommLock);
 
        for (request = requests; n > 0; request++, n--)
            RememberSyncRequest(&request->ftag, request->type);
 
        END_CRIT_SECTION();
    } while (loop);
 
    if (requests)
        pfree(requests);
}

References AmCheckpointerProcess, CheckpointerShmem, CKPT_REQ_BATCH_SIZE, END_CRIT_SECTION, fb(), CheckpointerShmemStruct::head, i, LW_EXCLUSIVE, LWLockAcquire(), LWLockRelease(), CheckpointerShmemStruct::max_requests, Min, CheckpointerShmemStruct::num_requests, palloc(), pfree(), RememberSyncRequest(), CheckpointerShmemStruct::requests, and START_CRIT_SECTION.

Referenced by CheckpointerMain(), CheckpointWriteDelay(), CreateCheckPoint(), ProcessSyncRequests(), SyncPostCheckpoint(), and SyncPreCheckpoint().

◆ BackgroundWriterMain()

pg_noreturn void BackgroundWriterMain	(	const void *	startup_data,
		size_t	startup_data_len
	)

extern

Definition at line 89 of file bgwriter.c.

{
    sigjmp_buf  local_sigjmp_buf;
    MemoryContext bgwriter_context;
    bool        prev_hibernate;
    WritebackContext wb_context;
 
    Assert(startup_data_len == 0);
 
    AuxiliaryProcessMainCommon();
 
    /*
     * Properly accept or ignore signals that might be sent to us.
     */
    pqsignal(SIGHUP, SignalHandlerForConfigReload);
    pqsignal(SIGINT, PG_SIG_IGN);
    pqsignal(SIGTERM, SignalHandlerForShutdownRequest);
    /* SIGQUIT handler was already set up by InitPostmasterChild */
    pqsignal(SIGALRM, PG_SIG_IGN);
    pqsignal(SIGPIPE, PG_SIG_IGN);
    pqsignal(SIGUSR1, procsignal_sigusr1_handler);
    pqsignal(SIGUSR2, PG_SIG_IGN);
 
    /*
     * Reset some signals that are accepted by postmaster but not here
     */
    pqsignal(SIGCHLD, PG_SIG_DFL);
 
    /*
     * We just started, assume there has been either a shutdown or
     * end-of-recovery snapshot.
     */
    last_snapshot_ts = GetCurrentTimestamp();
 
    /*
     * Create a memory context that we will do all our work in.  We do this so
     * that we can reset the context during error recovery and thereby avoid
     * possible memory leaks.  Formerly this code just ran in
     * TopMemoryContext, but resetting that would be a really bad idea.
     */
    bgwriter_context = AllocSetContextCreate(TopMemoryContext,
                                             "Background Writer",
                                             ALLOCSET_DEFAULT_SIZES);
    MemoryContextSwitchTo(bgwriter_context);
 
    WritebackContextInit(&wb_context, &bgwriter_flush_after);
 
    /*
     * If an exception is encountered, processing resumes here.
     *
     * You might wonder why this isn't coded as an infinite loop around a
     * PG_TRY construct.  The reason is that this is the bottom of the
     * exception stack, and so with PG_TRY there would be no exception handler
     * in force at all during the CATCH part.  By leaving the outermost setjmp
     * always active, we have at least some chance of recovering from an error
     * during error recovery.  (If we get into an infinite loop thereby, it
     * will soon be stopped by overflow of elog.c's internal state stack.)
     *
     * Note that we use sigsetjmp(..., 1), so that the prevailing signal mask
     * (to wit, BlockSig) will be restored when longjmp'ing to here.  Thus,
     * signals other than SIGQUIT will be blocked until we complete error
     * recovery.  It might seem that this policy makes the HOLD_INTERRUPTS()
     * call redundant, but it is not since InterruptPending might be set
     * already.
     */
    if (sigsetjmp(local_sigjmp_buf, 1) != 0)
    {
        /* Since not using PG_TRY, must reset error stack by hand */
        error_context_stack = NULL;
 
        /* Prevent interrupts while cleaning up */
        HOLD_INTERRUPTS();
 
        /* Report the error to the server log */
        EmitErrorReport();
 
        /*
         * These operations are really just a minimal subset of
         * AbortTransaction().  We don't have very many resources to worry
         * about in bgwriter, but we do have LWLocks, buffers, and temp files.
         */
        LWLockReleaseAll();
        ConditionVariableCancelSleep();
        pgaio_error_cleanup();
        UnlockBuffers();
        ReleaseAuxProcessResources(false);
        AtEOXact_Buffers(false);
        AtEOXact_SMgr();
        AtEOXact_Files(false);
        AtEOXact_HashTables(false);
 
        /*
         * Now return to normal top-level context and clear ErrorContext for
         * next time.
         */
        MemoryContextSwitchTo(bgwriter_context);
        FlushErrorState();
 
        /* Flush any leaked data in the top-level context */
        MemoryContextReset(bgwriter_context);
 
        /* re-initialize to avoid repeated errors causing problems */
        WritebackContextInit(&wb_context, &bgwriter_flush_after);
 
        /* Now we can allow interrupts again */
        RESUME_INTERRUPTS();
 
        /*
         * Sleep at least 1 second after any error.  A write error is likely
         * to be repeated, and we don't want to be filling the error logs as
         * fast as we can.
         */
        pg_usleep(1000000L);
 
        /* Report wait end here, when there is no further possibility of wait */
        pgstat_report_wait_end();
    }
 
    /* We can now handle ereport(ERROR) */
    PG_exception_stack = &local_sigjmp_buf;
 
    /*
     * Unblock signals (they were blocked when the postmaster forked us)
     */
    sigprocmask(SIG_SETMASK, &UnBlockSig, NULL);
 
    /*
     * Reset hibernation state after any error.
     */
    prev_hibernate = false;
 
    /*
     * Loop forever
     */
    for (;;)
    {
        bool        can_hibernate;
        int         rc;
 
        /* Clear any already-pending wakeups */
        ResetLatch(MyLatch);
 
        ProcessMainLoopInterrupts();
 
        /*
         * Do one cycle of dirty-buffer writing.
         */
        can_hibernate = BgBufferSync(&wb_context);
 
        /* Report pending statistics to the cumulative stats system */
        pgstat_report_bgwriter();
        pgstat_report_wal(true);
 
        if (FirstCallSinceLastCheckpoint())
        {
            /*
             * After any checkpoint, free all smgr objects.  Otherwise we
             * would never do so for dropped relations, as the bgwriter does
             * not process shared invalidation messages or call
             * AtEOXact_SMgr().
             */
            smgrdestroyall();
        }
 
        /*
         * Log a new xl_running_xacts every now and then so replication can
         * get into a consistent state faster (think of suboverflowed
         * snapshots) and clean up resources (locks, KnownXids*) more
         * frequently. The costs of this are relatively low, so doing it 4
         * times (LOG_SNAPSHOT_INTERVAL_MS) a minute seems fine.
         *
         * We assume the interval for writing xl_running_xacts is
         * significantly bigger than BgWriterDelay, so we don't complicate the
         * overall timeout handling but just assume we're going to get called
         * often enough even if hibernation mode is active. It's not that
         * important that LOG_SNAPSHOT_INTERVAL_MS is met strictly. To make
         * sure we're not waking the disk up unnecessarily on an idle system
         * we check whether there has been any WAL inserted since the last
         * time we've logged a running xacts.
         *
         * We do this logging in the bgwriter as it is the only process that
         * is run regularly and returns to its mainloop all the time. E.g.
         * Checkpointer, when active, is barely ever in its mainloop and thus
         * makes it hard to log regularly.
         */
        if (XLogStandbyInfoActive() && !RecoveryInProgress())
        {
            TimestampTz timeout = 0;
            TimestampTz now = GetCurrentTimestamp();
 
            timeout = TimestampTzPlusMilliseconds(last_snapshot_ts,
                                                  LOG_SNAPSHOT_INTERVAL_MS);
 
            /*
             * Only log if enough time has passed and interesting records have
             * been inserted since the last snapshot.  Have to compare with <=
             * instead of < because GetLastImportantRecPtr() points at the
             * start of a record, whereas last_snapshot_lsn points just past
             * the end of the record.
             */
            if (now >= timeout &&
                last_snapshot_lsn <= GetLastImportantRecPtr())
            {
                last_snapshot_lsn = LogStandbySnapshot();
                last_snapshot_ts = now;
            }
        }
 
        /*
         * Sleep until we are signaled or BgWriterDelay has elapsed.
         *
         * Note: the feedback control loop in BgBufferSync() expects that we
         * will call it every BgWriterDelay msec.  While it's not critical for
         * correctness that that be exact, the feedback loop might misbehave
         * if we stray too far from that.  Hence, avoid loading this process
         * down with latch events that are likely to happen frequently during
         * normal operation.
         */
        rc = WaitLatch(MyLatch,
                       WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH,
                       BgWriterDelay /* ms */ , WAIT_EVENT_BGWRITER_MAIN);
 
        /*
         * If no latch event and BgBufferSync says nothing's happening, extend
         * the sleep in "hibernation" mode, where we sleep for much longer
         * than bgwriter_delay says.  Fewer wakeups save electricity.  When a
         * backend starts using buffers again, it will wake us up by setting
         * our latch.  Because the extra sleep will persist only as long as no
         * buffer allocations happen, this should not distort the behavior of
         * BgBufferSync's control loop too badly; essentially, it will think
         * that the system-wide idle interval didn't exist.
         *
         * There is a race condition here, in that a backend might allocate a
         * buffer between the time BgBufferSync saw the alloc count as zero
         * and the time we call StrategyNotifyBgWriter.  While it's not
         * critical that we not hibernate anyway, we try to reduce the odds of
         * that by only hibernating when BgBufferSync says nothing's happening
         * for two consecutive cycles.  Also, we mitigate any possible
         * consequences of a missed wakeup by not hibernating forever.
         */
        if (rc == WL_TIMEOUT && can_hibernate && prev_hibernate)
        {
            /* Ask for notification at next buffer allocation */
            StrategyNotifyBgWriter(MyProcNumber);
            /* Sleep ... */
            (void) WaitLatch(MyLatch,
                             WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH,
                             BgWriterDelay * HIBERNATE_FACTOR,
                             WAIT_EVENT_BGWRITER_HIBERNATE);
            /* Reset the notification request in case we timed out */
            StrategyNotifyBgWriter(-1);
        }
 
        prev_hibernate = can_hibernate;
    }
}

References ALLOCSET_DEFAULT_SIZES, AllocSetContextCreate, Assert, AtEOXact_Buffers(), AtEOXact_Files(), AtEOXact_HashTables(), AtEOXact_SMgr(), AuxiliaryProcessMainCommon(), BgBufferSync(), bgwriter_flush_after, BgWriterDelay, ConditionVariableCancelSleep(), EmitErrorReport(), error_context_stack, fb(), FirstCallSinceLastCheckpoint(), FlushErrorState(), GetCurrentTimestamp(), GetLastImportantRecPtr(), HIBERNATE_FACTOR, HOLD_INTERRUPTS, last_snapshot_lsn, last_snapshot_ts, LOG_SNAPSHOT_INTERVAL_MS, LogStandbySnapshot(), LWLockReleaseAll(), MemoryContextReset(), MemoryContextSwitchTo(), MyLatch, MyProcNumber, now(), PG_exception_stack, PG_SIG_DFL, PG_SIG_IGN, pg_usleep(), pgaio_error_cleanup(), pgstat_report_bgwriter(), pgstat_report_wait_end(), pgstat_report_wal(), pqsignal, ProcessMainLoopInterrupts(), procsignal_sigusr1_handler(), RecoveryInProgress(), ReleaseAuxProcessResources(), ResetLatch(), RESUME_INTERRUPTS, SIGALRM, SIGCHLD, SIGHUP, SignalHandlerForConfigReload(), SignalHandlerForShutdownRequest(), SIGPIPE, SIGUSR1, SIGUSR2, smgrdestroyall(), StrategyNotifyBgWriter(), TimestampTzPlusMilliseconds, TopMemoryContext, UnBlockSig, UnlockBuffers(), WaitLatch(), WL_EXIT_ON_PM_DEATH, WL_LATCH_SET, WL_TIMEOUT, WritebackContextInit(), and XLogStandbyInfoActive.

◆ CheckpointerMain()

pg_noreturn void CheckpointerMain	(	const void *	startup_data,
		size_t	startup_data_len
	)

extern

Definition at line 206 of file checkpointer.c.

{
    sigjmp_buf  local_sigjmp_buf;
    MemoryContext checkpointer_context;
 
    Assert(startup_data_len == 0);
 
    AuxiliaryProcessMainCommon();
 
    CheckpointerShmem->checkpointer_pid = MyProcPid;
 
    /*
     * Properly accept or ignore signals the postmaster might send us
     *
     * Note: we deliberately ignore SIGTERM, because during a standard Unix
     * system shutdown cycle, init will SIGTERM all processes at once.  We
     * want to wait for the backends to exit, whereupon the postmaster will
     * tell us it's okay to shut down (via SIGUSR2).
     */
    pqsignal(SIGHUP, SignalHandlerForConfigReload);
    pqsignal(SIGINT, ReqShutdownXLOG);
    pqsignal(SIGTERM, PG_SIG_IGN);  /* ignore SIGTERM */
    /* SIGQUIT handler was already set up by InitPostmasterChild */
    pqsignal(SIGALRM, PG_SIG_IGN);
    pqsignal(SIGPIPE, PG_SIG_IGN);
    pqsignal(SIGUSR1, procsignal_sigusr1_handler);
    pqsignal(SIGUSR2, SignalHandlerForShutdownRequest);
 
    /*
     * Reset some signals that are accepted by postmaster but not here
     */
    pqsignal(SIGCHLD, PG_SIG_DFL);
 
    /*
     * Initialize so that first time-driven event happens at the correct time.
     */
    last_checkpoint_time = last_xlog_switch_time = (pg_time_t) time(NULL);
 
    /*
     * Write out stats after shutdown. This needs to be called by exactly one
     * process during a normal shutdown, and since checkpointer is shut down
     * very late...
     *
     * While e.g. walsenders are active after the shutdown checkpoint has been
     * written (and thus could produce more stats), checkpointer stays around
     * after the shutdown checkpoint has been written. postmaster will only
     * signal checkpointer to exit after all processes that could emit stats
     * have been shut down.
     */
    before_shmem_exit(pgstat_before_server_shutdown, 0);
 
    /*
     * Create a memory context that we will do all our work in.  We do this so
     * that we can reset the context during error recovery and thereby avoid
     * possible memory leaks.  Formerly this code just ran in
     * TopMemoryContext, but resetting that would be a really bad idea.
     */
    checkpointer_context = AllocSetContextCreate(TopMemoryContext,
                                                 "Checkpointer",
                                                 ALLOCSET_DEFAULT_SIZES);
    MemoryContextSwitchTo(checkpointer_context);
 
    /*
     * If an exception is encountered, processing resumes here.
     *
     * You might wonder why this isn't coded as an infinite loop around a
     * PG_TRY construct.  The reason is that this is the bottom of the
     * exception stack, and so with PG_TRY there would be no exception handler
     * in force at all during the CATCH part.  By leaving the outermost setjmp
     * always active, we have at least some chance of recovering from an error
     * during error recovery.  (If we get into an infinite loop thereby, it
     * will soon be stopped by overflow of elog.c's internal state stack.)
     *
     * Note that we use sigsetjmp(..., 1), so that the prevailing signal mask
     * (to wit, BlockSig) will be restored when longjmp'ing to here.  Thus,
     * signals other than SIGQUIT will be blocked until we complete error
     * recovery.  It might seem that this policy makes the HOLD_INTERRUPTS()
     * call redundant, but it is not since InterruptPending might be set
     * already.
     */
    if (sigsetjmp(local_sigjmp_buf, 1) != 0)
    {
        /* Since not using PG_TRY, must reset error stack by hand */
        error_context_stack = NULL;
 
        /* Prevent interrupts while cleaning up */
        HOLD_INTERRUPTS();
 
        /* Report the error to the server log */
        EmitErrorReport();
 
        /*
         * These operations are really just a minimal subset of
         * AbortTransaction().  We don't have very many resources to worry
         * about in checkpointer, but we do have LWLocks, buffers, and temp
         * files.
         */
        LWLockReleaseAll();
        ConditionVariableCancelSleep();
        pgstat_report_wait_end();
        pgaio_error_cleanup();
        UnlockBuffers();
        ReleaseAuxProcessResources(false);
        AtEOXact_Buffers(false);
        AtEOXact_SMgr();
        AtEOXact_Files(false);
        AtEOXact_HashTables(false);
 
        /* Warn any waiting backends that the checkpoint failed. */
        if (ckpt_active)
        {
            SpinLockAcquire(&CheckpointerShmem->ckpt_lck);
            CheckpointerShmem->ckpt_failed++;
            CheckpointerShmem->ckpt_done = CheckpointerShmem->ckpt_started;
            SpinLockRelease(&CheckpointerShmem->ckpt_lck);
 
            ConditionVariableBroadcast(&CheckpointerShmem->done_cv);
 
            ckpt_active = false;
        }
 
        /*
         * Now return to normal top-level context and clear ErrorContext for
         * next time.
         */
        MemoryContextSwitchTo(checkpointer_context);
        FlushErrorState();
 
        /* Flush any leaked data in the top-level context */
        MemoryContextReset(checkpointer_context);
 
        /* Now we can allow interrupts again */
        RESUME_INTERRUPTS();
 
        /*
         * Sleep at least 1 second after any error.  A write error is likely
         * to be repeated, and we don't want to be filling the error logs as
         * fast as we can.
         */
        pg_usleep(1000000L);
    }
 
    /* We can now handle ereport(ERROR) */
    PG_exception_stack = &local_sigjmp_buf;
 
    /*
     * Unblock signals (they were blocked when the postmaster forked us)
     */
    sigprocmask(SIG_SETMASK, &UnBlockSig, NULL);
 
    /*
     * Ensure all shared memory values are set correctly for the config. Doing
     * this here ensures no race conditions from other concurrent updaters.
     */
    UpdateSharedMemoryConfig();
 
    /*
     * Loop until we've been asked to write the shutdown checkpoint or
     * terminate.
     */
    for (;;)
    {
        bool        do_checkpoint = false;
        int         flags = 0;
        pg_time_t   now;
        int         elapsed_secs;
        int         cur_timeout;
        bool        chkpt_or_rstpt_requested = false;
        bool        chkpt_or_rstpt_timed = false;
 
        /* Clear any already-pending wakeups */
        ResetLatch(MyLatch);
 
        /*
         * Process any requests or signals received recently.
         */
        AbsorbSyncRequests();
 
        ProcessCheckpointerInterrupts();
        if (ShutdownXLOGPending || ShutdownRequestPending)
            break;
 
        /*
         * Detect a pending checkpoint request by checking whether the flags
         * word in shared memory is nonzero.  We shouldn't need to acquire the
         * ckpt_lck for this.
         */
        if (((volatile CheckpointerShmemStruct *) CheckpointerShmem)->ckpt_flags)
        {
            do_checkpoint = true;
            chkpt_or_rstpt_requested = true;
        }
 
        /*
         * Force a checkpoint if too much time has elapsed since the last one.
         * Note that we count a timed checkpoint in stats only when this
         * occurs without an external request, but we set the CAUSE_TIME flag
         * bit even if there is also an external request.
         */
        now = (pg_time_t) time(NULL);
        elapsed_secs = now - last_checkpoint_time;
        if (elapsed_secs >= CheckPointTimeout)
        {
            if (!do_checkpoint)
                chkpt_or_rstpt_timed = true;
            do_checkpoint = true;
            flags |= CHECKPOINT_CAUSE_TIME;
        }
 
        /*
         * Do a checkpoint if requested.
         */
        if (do_checkpoint)
        {
            bool        ckpt_performed = false;
            bool        do_restartpoint;
 
            /* Check if we should perform a checkpoint or a restartpoint. */
            do_restartpoint = RecoveryInProgress();
 
            /*
             * Atomically fetch the request flags to figure out what kind of a
             * checkpoint we should perform, and increase the started-counter
             * to acknowledge that we've started a new checkpoint.
             */
            SpinLockAcquire(&CheckpointerShmem->ckpt_lck);
            flags |= CheckpointerShmem->ckpt_flags;
            CheckpointerShmem->ckpt_flags = 0;
            CheckpointerShmem->ckpt_started++;
            SpinLockRelease(&CheckpointerShmem->ckpt_lck);
 
            ConditionVariableBroadcast(&CheckpointerShmem->start_cv);
 
            /*
             * The end-of-recovery checkpoint is a real checkpoint that's
             * performed while we're still in recovery.
             */
            if (flags & CHECKPOINT_END_OF_RECOVERY)
                do_restartpoint = false;
 
            if (chkpt_or_rstpt_timed)
            {
                chkpt_or_rstpt_timed = false;
                if (do_restartpoint)
                    PendingCheckpointerStats.restartpoints_timed++;
                else
                    PendingCheckpointerStats.num_timed++;
            }
 
            if (chkpt_or_rstpt_requested)
            {
                chkpt_or_rstpt_requested = false;
                if (do_restartpoint)
                    PendingCheckpointerStats.restartpoints_requested++;
                else
                    PendingCheckpointerStats.num_requested++;
            }
 
            /*
             * We will warn if (a) too soon since last checkpoint (whatever
             * caused it) and (b) somebody set the CHECKPOINT_CAUSE_XLOG flag
             * since the last checkpoint start.  Note in particular that this
             * implementation will not generate warnings caused by
             * CheckPointTimeout < CheckPointWarning.
             */
            if (!do_restartpoint &&
                (flags & CHECKPOINT_CAUSE_XLOG) &&
                elapsed_secs < CheckPointWarning)
                ereport(LOG,
                        (errmsg_plural("checkpoints are occurring too frequently (%d second apart)",
                                       "checkpoints are occurring too frequently (%d seconds apart)",
                                       elapsed_secs,
                                       elapsed_secs),
                         errhint("Consider increasing the configuration parameter \"%s\".", "max_wal_size")));
 
            /*
             * Initialize checkpointer-private variables used during
             * checkpoint.
             */
            ckpt_active = true;
            if (do_restartpoint)
                ckpt_start_recptr = GetXLogReplayRecPtr(NULL);
            else
                ckpt_start_recptr = GetInsertRecPtr();
            ckpt_start_time = now;
            ckpt_cached_elapsed = 0;
 
            /*
             * Do the checkpoint.
             */
            if (!do_restartpoint)
                ckpt_performed = CreateCheckPoint(flags);
            else
                ckpt_performed = CreateRestartPoint(flags);
 
            /*
             * After any checkpoint, free all smgr objects.  Otherwise we
             * would never do so for dropped relations, as the checkpointer
             * does not process shared invalidation messages or call
             * AtEOXact_SMgr().
             */
            smgrdestroyall();
 
            /*
             * Indicate checkpoint completion to any waiting backends.
             */
            SpinLockAcquire(&CheckpointerShmem->ckpt_lck);
            CheckpointerShmem->ckpt_done = CheckpointerShmem->ckpt_started;
            SpinLockRelease(&CheckpointerShmem->ckpt_lck);
 
            ConditionVariableBroadcast(&CheckpointerShmem->done_cv);
 
            if (!do_restartpoint)
            {
                /*
                 * Note we record the checkpoint start time not end time as
                 * last_checkpoint_time.  This is so that time-driven
                 * checkpoints happen at a predictable spacing.
                 */
                last_checkpoint_time = now;
 
                if (ckpt_performed)
                    PendingCheckpointerStats.num_performed++;
            }
            else
            {
                if (ckpt_performed)
                {
                    /*
                     * The same as for checkpoint. Please see the
                     * corresponding comment.
                     */
                    last_checkpoint_time = now;
 
                    PendingCheckpointerStats.restartpoints_performed++;
                }
                else
                {
                    /*
                     * We were not able to perform the restartpoint
                     * (checkpoints throw an ERROR in case of error).  Most
                     * likely because we have not received any new checkpoint
                     * WAL records since the last restartpoint. Try again in
                     * 15 s.
                     */
                    last_checkpoint_time = now - CheckPointTimeout + 15;
                }
            }
 
            ckpt_active = false;
 
            /*
             * We may have received an interrupt during the checkpoint and the
             * latch might have been reset (e.g. in CheckpointWriteDelay).
             */
            ProcessCheckpointerInterrupts();
            if (ShutdownXLOGPending || ShutdownRequestPending)
                break;
        }
 
        /*
         * Disable logical decoding if someone requested it. See comments atop
         * logicalctl.c.
         */
        DisableLogicalDecodingIfNecessary();
 
        /* Check for archive_timeout and switch xlog files if necessary. */
        CheckArchiveTimeout();
 
        /* Report pending statistics to the cumulative stats system */
        pgstat_report_checkpointer();
        pgstat_report_wal(true);
 
        /*
         * If any checkpoint flags have been set, redo the loop to handle the
         * checkpoint without sleeping.
         */
        if (((volatile CheckpointerShmemStruct *) CheckpointerShmem)->ckpt_flags)
            continue;
 
        /*
         * Sleep until we are signaled or it's time for another checkpoint or
         * xlog file switch.
         */
        now = (pg_time_t) time(NULL);
        elapsed_secs = now - last_checkpoint_time;
        if (elapsed_secs >= CheckPointTimeout)
            continue;           /* no sleep for us ... */
        cur_timeout = CheckPointTimeout - elapsed_secs;
        if (XLogArchiveTimeout > 0 && !RecoveryInProgress())
        {
            elapsed_secs = now - last_xlog_switch_time;
            if (elapsed_secs >= XLogArchiveTimeout)
                continue;       /* no sleep for us ... */
            cur_timeout = Min(cur_timeout, XLogArchiveTimeout - elapsed_secs);
        }
 
        (void) WaitLatch(MyLatch,
                         WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH,
                         cur_timeout * 1000L /* convert to ms */ ,
                         WAIT_EVENT_CHECKPOINTER_MAIN);
    }
 
    /*
     * From here on, elog(ERROR) should end with exit(1), not send control
     * back to the sigsetjmp block above.
     */
    ExitOnAnyError = true;
 
    if (ShutdownXLOGPending)
    {
        /*
         * Close down the database.
         *
         * Since ShutdownXLOG() creates restartpoint or checkpoint, and
         * updates the statistics, increment the checkpoint request and flush
         * out pending statistic.
         */
        PendingCheckpointerStats.num_requested++;
        ShutdownXLOG(0, 0);
        pgstat_report_checkpointer();
        pgstat_report_wal(true);
 
        /*
         * Tell postmaster that we're done.
         */
        SendPostmasterSignal(PMSIGNAL_XLOG_IS_SHUTDOWN);
        ShutdownXLOGPending = false;
    }
 
    /*
     * Wait until we're asked to shut down. By separating the writing of the
     * shutdown checkpoint from checkpointer exiting, checkpointer can perform
     * some should-be-as-late-as-possible work like writing out stats.
     */
    for (;;)
    {
        /* Clear any already-pending wakeups */
        ResetLatch(MyLatch);
 
        ProcessCheckpointerInterrupts();
 
        if (ShutdownRequestPending)
            break;
 
        (void) WaitLatch(MyLatch,
                         WL_LATCH_SET | WL_EXIT_ON_PM_DEATH,
                         0,
                         WAIT_EVENT_CHECKPOINTER_SHUTDOWN);
    }
 
    /* Normal exit from the checkpointer is here */
    proc_exit(0);               /* done */
}

◆ CheckpointWriteDelay()

void CheckpointWriteDelay	(	int	flags,
		double	progress
	)

extern

Definition at line 795 of file checkpointer.c.

{
    static int  absorb_counter = WRITES_PER_ABSORB;
 
    /* Do nothing if checkpoint is being executed by non-checkpointer process */
    if (!AmCheckpointerProcess())
        return;
 
    /*
     * Perform the usual duties and take a nap, unless we're behind schedule,
     * in which case we just try to catch up as quickly as possible.
     */
    if (!(flags & CHECKPOINT_FAST) &&
        !ShutdownXLOGPending &&
        !ShutdownRequestPending &&
        !FastCheckpointRequested() &&
        IsCheckpointOnSchedule(progress))
    {
        if (ConfigReloadPending)
        {
            ConfigReloadPending = false;
            ProcessConfigFile(PGC_SIGHUP);
            /* update shmem copies of config variables */
            UpdateSharedMemoryConfig();
        }
 
        AbsorbSyncRequests();
        absorb_counter = WRITES_PER_ABSORB;
 
        CheckArchiveTimeout();
 
        /* Report interim statistics to the cumulative stats system */
        pgstat_report_checkpointer();
 
        /*
         * This sleep used to be connected to bgwriter_delay, typically 200ms.
         * That resulted in more frequent wakeups if not much work to do.
         * Checkpointer and bgwriter are no longer related so take the Big
         * Sleep.
         */
        WaitLatch(MyLatch, WL_LATCH_SET | WL_EXIT_ON_PM_DEATH | WL_TIMEOUT,
                  100,
                  WAIT_EVENT_CHECKPOINT_WRITE_DELAY);
        ResetLatch(MyLatch);
    }
    else if (--absorb_counter <= 0)
    {
        /*
         * Absorb pending fsync requests after each WRITES_PER_ABSORB write
         * operations even when we don't sleep, to prevent overflow of the
         * fsync request queue.
         */
        AbsorbSyncRequests();
        absorb_counter = WRITES_PER_ABSORB;
    }
 
    /* Check for barrier events. */
    if (ProcSignalBarrierPending)
        ProcessProcSignalBarrier();
}

References AbsorbSyncRequests(), AmCheckpointerProcess, CheckArchiveTimeout(), CHECKPOINT_FAST, ConfigReloadPending, FastCheckpointRequested(), fb(), IsCheckpointOnSchedule(), MyLatch, PGC_SIGHUP, pgstat_report_checkpointer(), ProcessConfigFile(), ProcessProcSignalBarrier(), ProcSignalBarrierPending, progress, ResetLatch(), ShutdownRequestPending, ShutdownXLOGPending, UpdateSharedMemoryConfig(), WaitLatch(), WL_EXIT_ON_PM_DEATH, WL_LATCH_SET, WL_TIMEOUT, and WRITES_PER_ABSORB.

Referenced by BufferSync().

◆ ExecCheckpoint()

void ExecCheckpoint	(	ParseState *	pstate,
		CheckPointStmt *	stmt
	)

extern

Definition at line 1001 of file checkpointer.c.

{
    bool        fast = true;
    bool        unlogged = false;
 
    foreach_ptr(DefElem, opt, stmt->options)
    {
        if (strcmp(opt->defname, "mode") == 0)
        {
            char       *mode = defGetString(opt);
 
            if (strcmp(mode, "spread") == 0)
                fast = false;
            else if (strcmp(mode, "fast") != 0)
                ereport(ERROR,
                        (errcode(ERRCODE_SYNTAX_ERROR),
                         errmsg("unrecognized value for %s option \"%s\": \"%s\"",
                                "CHECKPOINT", "mode", mode),
                         parser_errposition(pstate, opt->location)));
        }
        else if (strcmp(opt->defname, "flush_unlogged") == 0)
            unlogged = defGetBoolean(opt);
        else
            ereport(ERROR,
                    (errcode(ERRCODE_SYNTAX_ERROR),
                     errmsg("unrecognized %s option \"%s\"",
                            "CHECKPOINT", opt->defname),
                     parser_errposition(pstate, opt->location)));
    }
 
    if (!has_privs_of_role(GetUserId(), ROLE_PG_CHECKPOINT))
        ereport(ERROR,
                (errcode(ERRCODE_INSUFFICIENT_PRIVILEGE),
        /* translator: %s is name of an SQL command (e.g., CHECKPOINT) */
                 errmsg("permission denied to execute %s command",
                        "CHECKPOINT"),
                 errdetail("Only roles with privileges of the \"%s\" role may execute this command.",
                           "pg_checkpoint")));
 
    RequestCheckpoint(CHECKPOINT_WAIT |
                      (fast ? CHECKPOINT_FAST : 0) |
                      (unlogged ? CHECKPOINT_FLUSH_UNLOGGED : 0) |
                      (RecoveryInProgress() ? 0 : CHECKPOINT_FORCE));
}

References CHECKPOINT_FAST, CHECKPOINT_FLUSH_UNLOGGED, CHECKPOINT_FORCE, CHECKPOINT_WAIT, defGetBoolean(), defGetString(), ereport, errcode(), errdetail(), errmsg, ERROR, fb(), foreach_ptr, GetUserId(), has_privs_of_role(), mode, parser_errposition(), RecoveryInProgress(), RequestCheckpoint(), and stmt.

Referenced by standard_ProcessUtility().

◆ FirstCallSinceLastCheckpoint()

bool FirstCallSinceLastCheckpoint ( void )

extern

Definition at line 1514 of file checkpointer.c.

{
    static int  ckpt_done = 0;
    int         new_done;
    bool        FirstCall = false;
 
    SpinLockAcquire(&CheckpointerShmem->ckpt_lck);
    new_done = CheckpointerShmem->ckpt_done;
    SpinLockRelease(&CheckpointerShmem->ckpt_lck);
 
    if (new_done != ckpt_done)
        FirstCall = true;
 
    ckpt_done = new_done;
 
    return FirstCall;
}

References CheckpointerShmem, CheckpointerShmemStruct::ckpt_done, CheckpointerShmemStruct::ckpt_lck, fb(), SpinLockAcquire(), and SpinLockRelease().

Referenced by BackgroundWriterMain().

◆ ForwardSyncRequest()

bool ForwardSyncRequest	(	const FileTag *	ftag,
		SyncRequestType	type
	)

extern

Definition at line 1213 of file checkpointer.c.

{
    CheckpointerRequest *request;
    bool        too_full;
    int         insert_pos;
 
    if (!IsUnderPostmaster)
        return false;           /* probably shouldn't even get here */
 
    if (AmCheckpointerProcess())
        elog(ERROR, "ForwardSyncRequest must not be called in checkpointer");
 
    LWLockAcquire(CheckpointerCommLock, LW_EXCLUSIVE);
 
    /*
     * If the checkpointer isn't running or the request queue is full, the
     * backend will have to perform its own fsync request.  But before forcing
     * that to happen, we can try to compact the request queue.
     */
    if (CheckpointerShmem->checkpointer_pid == 0 ||
        (CheckpointerShmem->num_requests >= CheckpointerShmem->max_requests &&
         !CompactCheckpointerRequestQueue()))
    {
        LWLockRelease(CheckpointerCommLock);
        return false;
    }
 
    /* OK, insert request */
    insert_pos = CheckpointerShmem->tail;
    request = &CheckpointerShmem->requests[insert_pos];
    request->ftag = *ftag;
    request->type = type;
 
    CheckpointerShmem->tail = (CheckpointerShmem->tail + 1) % CheckpointerShmem->max_requests;
    CheckpointerShmem->num_requests++;
 
    /* If queue is more than half full, nudge the checkpointer to empty it */
    too_full = (CheckpointerShmem->num_requests >=
                CheckpointerShmem->max_requests / 2);
 
    LWLockRelease(CheckpointerCommLock);
 
    /* ... but not till after we release the lock */
    if (too_full)
    {
        ProcNumber  checkpointerProc = pg_atomic_read_u32(&ProcGlobal->checkpointerProc);
 
        if (checkpointerProc != INVALID_PROC_NUMBER)
            SetLatch(&GetPGProcByNumber(checkpointerProc)->procLatch);
    }
 
    return true;
}

References AmCheckpointerProcess, CheckpointerShmemStruct::checkpointer_pid, PROC_HDR::checkpointerProc, CheckpointerShmem, CompactCheckpointerRequestQueue(), elog, ERROR, fb(), CheckpointerRequest::ftag, GetPGProcByNumber, INVALID_PROC_NUMBER, IsUnderPostmaster, LW_EXCLUSIVE, LWLockAcquire(), LWLockRelease(), CheckpointerShmemStruct::max_requests, CheckpointerShmemStruct::num_requests, pg_atomic_read_u32(), ProcGlobal, CheckpointerShmemStruct::requests, SetLatch(), CheckpointerShmemStruct::tail, and type.

Referenced by RegisterSyncRequest().

◆ RequestCheckpoint()

void RequestCheckpoint ( int flags )

extern

Definition at line 1064 of file checkpointer.c.

{
    int         ntries;
    int         old_failed,
                old_started;
 
    /*
     * If in a standalone backend, just do it ourselves.
     */
    if (!IsPostmasterEnvironment)
    {
        /*
         * There's no point in doing slow checkpoints in a standalone backend,
         * because there's no other backends the checkpoint could disrupt.
         */
        CreateCheckPoint(flags | CHECKPOINT_FAST);
 
        /* Free all smgr objects, as CheckpointerMain() normally would. */
        smgrdestroyall();
 
        return;
    }
 
    /*
     * Atomically set the request flags, and take a snapshot of the counters.
     * When we see ckpt_started > old_started, we know the flags we set here
     * have been seen by checkpointer.
     *
     * Note that we OR the flags with any existing flags, to avoid overriding
     * a "stronger" request by another backend.  The flag senses must be
     * chosen to make this work!
     */
    SpinLockAcquire(&CheckpointerShmem->ckpt_lck);
 
    old_failed = CheckpointerShmem->ckpt_failed;
    old_started = CheckpointerShmem->ckpt_started;
    CheckpointerShmem->ckpt_flags |= (flags | CHECKPOINT_REQUESTED);
 
    SpinLockRelease(&CheckpointerShmem->ckpt_lck);
 
    /*
     * Set checkpointer's latch to request checkpoint.  It's possible that the
     * checkpointer hasn't started yet, so we will retry a few times if
     * needed.  (Actually, more than a few times, since on slow or overloaded
     * buildfarm machines, it's been observed that the checkpointer can take
     * several seconds to start.)  However, if not told to wait for the
     * checkpoint to occur, we consider failure to set the latch to be
     * nonfatal and merely LOG it.  The checkpointer should see the request
     * when it does start, with or without the SetLatch().
     */
#define MAX_SIGNAL_TRIES 600    /* max wait 60.0 sec */
    for (ntries = 0;; ntries++)
    {
        ProcNumber  checkpointerProc = pg_atomic_read_u32(&ProcGlobal->checkpointerProc);
 
        if (checkpointerProc == INVALID_PROC_NUMBER)
        {
            if (ntries >= MAX_SIGNAL_TRIES || !(flags & CHECKPOINT_WAIT))
            {
                elog((flags & CHECKPOINT_WAIT) ? ERROR : LOG,
                     "could not notify checkpoint: checkpointer is not running");
                break;
            }
        }
        else
        {
            SetLatch(&GetPGProcByNumber(checkpointerProc)->procLatch);
            /* notified successfully */
            break;
        }
 
        CHECK_FOR_INTERRUPTS();
        pg_usleep(100000L);     /* wait 0.1 sec, then retry */
    }
 
    /*
     * If requested, wait for completion.  We detect completion according to
     * the algorithm given above.
     */
    if (flags & CHECKPOINT_WAIT)
    {
        int         new_started,
                    new_failed;
 
        /* Wait for a new checkpoint to start. */
        ConditionVariablePrepareToSleep(&CheckpointerShmem->start_cv);
        for (;;)
        {
            SpinLockAcquire(&CheckpointerShmem->ckpt_lck);
            new_started = CheckpointerShmem->ckpt_started;
            SpinLockRelease(&CheckpointerShmem->ckpt_lck);
 
            if (new_started != old_started)
                break;
 
            ConditionVariableSleep(&CheckpointerShmem->start_cv,
                                   WAIT_EVENT_CHECKPOINT_START);
        }
        ConditionVariableCancelSleep();
 
        /*
         * We are waiting for ckpt_done >= new_started, in a modulo sense.
         */
        ConditionVariablePrepareToSleep(&CheckpointerShmem->done_cv);
        for (;;)
        {
            int         new_done;
 
            SpinLockAcquire(&CheckpointerShmem->ckpt_lck);
            new_done = CheckpointerShmem->ckpt_done;
            new_failed = CheckpointerShmem->ckpt_failed;
            SpinLockRelease(&CheckpointerShmem->ckpt_lck);
 
            if (new_done - new_started >= 0)
                break;
 
            ConditionVariableSleep(&CheckpointerShmem->done_cv,
                                   WAIT_EVENT_CHECKPOINT_DONE);
        }
        ConditionVariableCancelSleep();
 
        if (new_failed != old_failed)
            ereport(ERROR,
                    (errmsg("checkpoint request failed"),
                     errhint("Consult recent messages in the server log for details.")));
    }
}