PG起库逻辑和spill导致起库慢问题分析

/*
 * Given a replication slot, transaction ID and segment number, fill in the
 * corresponding spill file into 'path', which is a caller-owned buffer of size
 * at least MAXPGPATH.
 */
static void
ReorderBufferSerializedPath(char *path, ReplicationSlot *slot, TransactionId xid,
                            XLogSegNo segno)
{
    XLogRecPtr    recptr;

    XLogSegNoOffsetToRecPtr(segno, 0, wal_segment_size, recptr);

    snprintf(path, MAXPGPATH, "pg_replslot/%s/xid-%u-lsn-%X-%X.spill",
             NameStr(MyReplicationSlot->data.name),
             xid,
             (uint32) (recptr >> 32), (uint32) recptr);
}

/*
 * Pointer to a location in the XLOG.  These pointers are 64 bits wide,
 * because we don't want them ever to overflow.
 */
typedef uint64 XLogRecPtr;

#define XLogSegNoOffsetToRecPtr(segno, offset, wal_segsz_bytes, dest) \
        (dest) = (segno) * (wal_segsz_bytes) + (offset)

segno= dest/(128*1024*1024)
-- 再把16进制20000000转化下
segno= x'20000000'::int/(128*1024*1024)
segno= 4

ls 42D1E*04
0000000200042D1E00000004

## 恢复出的环境

[postgres]$ ll |head -100
total 40000276
-rw------- 1 postgres postgres  184 Dec  6 15:20 state
-rw------- 1 postgres postgres  196 Dec  6 13:25 xid-407989064-lsn-42D1E-0.spill
-rw------- 1 postgres postgres  208 Dec  6 13:25 xid-407989064-lsn-42D1E-20000000.spill
...
-rw------- 1 postgres postgres  540 Dec  6 16:44 xid-407989064-lsn-42D2A-D0000000.spill
-rw------- 1 postgres postgres  152 Dec  6 13:09 xid-407989065-lsn-42D1D-C8000000.spill
-rw------- 1 postgres postgres  152 Dec  6 13:09 xid-407989066-lsn-42D1D-C8000000.spill
-rw------- 1 postgres postgres  152 Dec  6 13:09 xid-407989068-lsn-42D1D-C8000000.spill
-rw------- 1 postgres postgres  152 Dec  6 13:09 xid-407989070-lsn-42D1D-C8000000.spill
-rw------- 1 postgres postgres  152 Dec  6 13:09 xid-407989072-lsn-42D1D-C8000000.spill
-rw------- 1 postgres postgres  152 Dec  6 13:09 xid-407989076-lsn-42D1D-C8000000.spill
-rw------- 1 postgres postgres  152 Dec  6 13:09 xid-407989079-lsn-42D1D-C8000000.spill
-rw------- 1 postgres postgres  152 Dec  6 13:09 xid-407989080-lsn-42D1D-C8000000.spill
-rw------- 1 postgres postgres  152 Dec  6 13:09 xid-407989082-lsn-42D1D-C8000000.spill

[postgres@lzlhost /myhost/liuzhilong/pg_replslot/logical_ex9e15_rep]$ ll |awk '{print $9}'|awk -F '-' '{print $2}'|sort|uniq -c|wc -l
10000003
[postgres@lzlhost /myhost/liuzhilong/pg_replslot/logical_ex9e15_rep]$ ll |wc -l
10000070

--发布订阅搭建复制链路
logical_decoding_work_mem = 64MB   #pg_ctl reload
wal_segment_size =128 MB

--source
CREATE TABLE replication_table (
    id BIGSERIAL PRIMARY KEY,
    column1 char(2000),
    column2 char(2000),
    column3 char(2000)
);
create publication pub_test for table replication_table ;

--dest
CREATE TABLE replication_table (
    id BIGSERIAL PRIMARY KEY,
    column1 char(2000),
    column2 char(2000),
    column3 char(2000)
);

CREATE SUBSCRIPTION sub_test
CONNECTION 'host=127.0.0.1 port=8094 dbname=lzl user=lzl password=qwer'
PUBLICATION pub_test;

--source
select * from pg_replication_slots;

--创建一个大事务暂时不提交
begin;
insert into replication_table(column1,column2,column3) 
select 'a','b','c' from generate_series(1,1000000) g;

--复制槽溢出
 ll
total 331924
-rw------- 1 postgres postgres       184 Dec  9 20:22 state
-rw------- 1 postgres postgres  88226964 Dec  9 20:22 xid-5074343-lsn-163-38000000.spill
-rw------- 1 postgres postgres 119698488 Dec  9 20:22 xid-5074343-lsn-163-40000000.spill

M=# select pid,usename,sent_lsn,write_lsn,flush_lsn,replay_lsn,write_lag,flush_lag,replay_lag,reply_time from pg_stat_replication;
  pid   | usename |   sent_lsn   |  write_lsn   |  flush_lsn   |  replay_lsn  | write_lag | flush_lag | replay_lag |          reply_time          
--------+---------+--------------+--------------+--------------+--------------+-----------+-----------+------------+------------------------------
 163525 | lzl     | 163/4996E1C8 | 163/4996E1C8 | 163/4996E1C8 | 163/4996E1C8 | [null]    | [null]    | [null]     | 2024-12-09 20:25:35.14769+08
(1 row)

M=# select pid,usename,pg_wal_lsn_diff(pg_current_wal_lsn(),sent_lsn) diff_sent_mb,pg_wal_lsn_diff(pg_current_wal_lsn(),write_lsn) diff_write_mb,pg_wal_lsn_diff(pg_current_wal_lsn(),flush_lsn) diff_flush_mb,pg_wal_lsn_diff(pg_current_wal_lsn(),replay_lsn) diff_replay_mb,pg_walfile_name_offset(sent_lsn) sentoffset,pg_walfile_name_offset(write_lsn) writeoffset,pg_walfile_name_offset(flush_lsn) flush_lsn from pg_stat_replication;
  pid   | usename | diff_sent_mb | diff_write_mb | diff_flush_mb | diff_replay_mb |             sentoffset              |             writeoffset             |              flush_lsn        
--------+---------+--------------+---------------+---------------+----------------+-------------------------------------+-------------------------------------+-------------------------------
 163525 | lzl     |            0 |             0 |             0 |              0 | (000000010000016300000009,26665416) | (000000010000016300000009,26665416) | (000000


[/mypg/pg8094/data/pg_replslot/sub_test]$ ll
total 357392
-rw------- 1 postgres postgres       184 Dec  9 20:23 state
-rw------- 1 postgres postgres  88226964 Dec  9 20:22 xid-5074343-lsn-163-38000000.spill
-rw------- 1 postgres postgres 137696328 Dec  9 20:23 xid-5074343-lsn-163-40000000.spill
-rw------- 1 postgres postgres  26076708 Dec  9 20:23 xid-5074343-lsn-163-48000000.spill
[/mypg/pg8094/data/pg_replslot/sub_test]$ ll
total 4
-rw------- 1 postgres postgres 184 Dec  9 20:25 state2666
(1 row)

--source 创建一个不相干的表，准备写入数据
CREATE TABLE no_replication_table (
    id BIGSERIAL PRIMARY KEY,
    column1 char(2000),
    column2 char(2000),
    column3 char(2000)
);

--创建一个大事务暂时不提交
begin;
insert into no_replication_table(column1,column2,column3) 
select 'a','b','c' from generate_series(1,1000000) g;

--溢出
[postgres@lzldb:MYINST:8094 /mypg/pg8094/data/pg_replslot/sub_test]$ ll
total 357492
-rw------- 1 postgres postgres       184 Dec  9 20:09 state
-rw------- 1 postgres postgres 107511456 Dec  9 20:08 xid-5074106-lsn-163-28000000.spill
-rw------- 1 postgres postgres 137698804 Dec  9 20:09 xid-5074106-lsn-163-30000000.spill
-rw------- 1 postgres postgres   4308444 Dec  9 20:09 xid-5074106-lsn-163-38000000.spill

## 一次insert一行，每个insert一个子事务
echo "begin;">subtx.sql
for i in {1..1000000}
do
  echo "savepoint p$i;">>subtx.sql
  echo "insert into no_replication_table(column1,column2,column3) select 'a','b','c';">>subtx.sql
done


nohup psql -d lzl -f subtx.sql &

#执行过程中，观察到溢出80w+文件
[/myhost/pg8094/data/pg_replslot/sub_test]$ ll |wc -l
823749
[/myhost/pg8094/data/pg_replslot/sub_test]$ ll |head -10
total 1099532
-rw------- 1 postgres postgres  184 Dec  9 21:10 state
-rw------- 1 postgres postgres 1236 Dec  9 21:10 xid-5519686-lsn-163-70000000.spill
-rw------- 1 postgres postgres  252 Dec  9 21:09 xid-5519687-lsn-163-70000000.spill
-rw------- 1 postgres postgres  252 Dec  9 21:09 xid-5519688-lsn-163-70000000.spill
-rw------- 1 postgres postgres  252 Dec  9 21:09 xid-5519689-lsn-163-70000000.spill
-rw------- 1 postgres postgres  252 Dec  9 21:09 xid-5519690-lsn-163-70000000.spill
-rw------- 1 postgres postgres  252 Dec  9 21:09 xid-5519691-lsn-163-70000000.spill
-rw------- 1 postgres postgres  252 Dec  9 21:09 xid-5519692-lsn-163-70000000.spill
-rw------- 1 postgres postgres  252 Dec  9 21:09 xid-5519693-lsn-163-70000000.spill

#define StartupDataBase()		StartChildProcess(StartupProcess)

	switch (MyAuxProcType)
	{
		case CheckerProcess:
			/* don't set signals, they're useless here */
			CheckerModeMain();
			proc_exit(1);		/* should never return */

		case BootstrapProcess:

			/*
			 * There was a brief instant during which mode was Normal; this is
			 * okay.  We need to be in bootstrap mode during BootStrapXLOG for
			 * the sake of multixact initialization.
			 */
			SetProcessingMode(BootstrapProcessing);
			bootstrap_signals();
			BootStrapXLOG();
			BootstrapModeMain();
			proc_exit(1);		/* should never return */

		case StartupProcess:   //这里这里这里这里
			/* don't set signals, startup process has its own agenda */
			StartupProcessMain();
			proc_exit(1);		/* should never return */

		case BgWriterProcess:
			/* don't set signals, bgwriter has its own agenda */
			BackgroundWriterMain();
			proc_exit(1);		/* should never return */

		case CheckpointerProcess:
			/* don't set signals, checkpointer has its own agenda */
			CheckpointerMain();
			proc_exit(1);		/* should never return */

		case WalWriterProcess:
			/* don't set signals, walwriter has its own agenda */
			InitXLOGAccess();
			WalWriterMain();
			proc_exit(1);		/* should never return */

		case WalReceiverProcess:
			/* don't set signals, walreceiver has its own agenda */
			WalReceiverMain();
			proc_exit(1);		/* should never return */

		default:
			elog(PANIC, "unrecognized process type: %d", (int) MyAuxProcType);
			proc_exit(1);
	}

This must be called ONCE during postmaster or standalone-backend startup

	switch (ControlFile->state)
	{
            ...
			case DB_IN_PRODUCTION:
			ereport(LOG,
					(errmsg("database system was interrupted; last known up at %s",
							str_time(ControlFile->time))));
			break;

2024-12-06 17:02:57.534 CST,,,447560,,65693cde.6d448,1325,,2023-12-01 09:54:38 CST,,0,LOG,00000,"database system is shut down",,,,,,,,,"","postmaster"
2024-12-06 17:03:49.536 CST,,,211844,,6752bdf3.33b84,1,,2024-12-06 17:03:47 CST,,0,LOG,00000,"ending log output to stderr",,"Future log output will go to log destination ""csvlog"".",,,,,,,"","postmaster"
2024-12-06 17:03:49.536 CST,,,211844,,6752bdf3.33b84,2,,2024-12-06 17:03:47 CST,,0,LOG,00000,"starting PostgreSQL 13.2 (RaseSQL 1.3) on x86_64-pc-linux-gnu, compiled by gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-39.0.1), 64-bit",,,,,,,,,"","postmaster"
2024-12-06 17:03:49.537 CST,,,211844,,6752bdf3.33b84,3,,2024-12-06 17:03:47 CST,,0,LOG,00000,"listening on IPv4 address ""0.0.0.0"", port 7284",,,,,,,,,"","postmaster"
2024-12-06 17:03:49.539 CST,,,211844,,6752bdf3.33b84,4,,2024-12-06 17:03:47 CST,,0,LOG,00000,"listening on Unix socket ""/tmp/.s.PGSQL.7284""",,,,,,,,,"","postmaster"
2024-12-06 17:03:49.557 CST,,,211995,,6752bdf5.33c1b,1,,2024-12-06 17:03:49 CST,,0,LOG,00000,"database system was interrupted; last known up at 2024-12-06 17:00:10 CST",,,,,,,,,"","startup"

Database cluster state:               in production

	/*----------
	 * If we previously crashed, perform a couple of actions:
	 *
	 * - The pg_wal directory may still include some temporary WAL segments
	 *   used when creating a new segment, so perform some clean up to not
	 *   bloat this path.  This is done first as there is no point to sync
	 *   this temporary data.
	 *
	 * - There might be data which we had written, intending to fsync it, but
	 *   which we had not actually fsync'd yet.  Therefore, a power failure in
	 *   the near future might cause earlier unflushed writes to be lost, even
	 *   though more recent data written to disk from here on would be
	 *   persisted.  To avoid that, fsync the entire data directory.
	 */
	if (ControlFile->state != DB_SHUTDOWNED &&
		ControlFile->state != DB_SHUTDOWNED_IN_RECOVERY)
	{
		RemoveTempXlogFiles();
		SyncDataDirectory();
	}

	/*
	 * Initialize replication slots, before there's a chance to remove
	 * required resources.
	 */
	StartupReplicationSlots();

	/*
	 * Startup logical state, needs to be setup now so we have proper data
	 * during crash recovery.
	 */
	StartupReorderBuffer();

/*
 * Delete all data spilled to disk after we've restarted/crashed. It will be
 * recreated when the respective slots are reused.
 */
void
StartupReorderBuffer(void)
{
	DIR		   *logical_dir;
	struct dirent *logical_de;

	logical_dir = AllocateDir("pg_replslot");
	while ((logical_de = ReadDir(logical_dir, "pg_replslot")) != NULL)
	{
		if (strcmp(logical_de->d_name, ".") == 0 ||
			strcmp(logical_de->d_name, "..") == 0)
			continue;

		/* if it cannot be a slot, skip the directory */
		if (!ReplicationSlotValidateName(logical_de->d_name, DEBUG2))
			continue;

		/*
		 * ok, has to be a surviving logical slot, iterate and delete
		 * everything starting with xid-*
		 */
		ReorderBufferCleanupSerializedTXNs(logical_de->d_name);
	}
	FreeDir(logical_dir);
}

/*
 * Issue fsync recursively on PGDATA and all its contents.
 *
 * We fsync regular files and directories wherever they are, but we
 * follow symlinks only for pg_wal and immediately under pg_tblspc.
 * Other symlinks are presumed to point at files we're not responsible
 * for fsyncing, and might not have privileges to write at all.
 *
 * Errors are logged but not considered fatal; that's because this is used
 * only during database startup, to deal with the possibility that there are
 * issued-but-unsynced writes pending against the data directory.  We want to
 * ensure that such writes reach disk before anything that's done in the new
 * run.  However, aborting on error would result in failure to start for
 * harmless cases such as read-only files in the data directory, and that's
 * not good either.
 *
 * Note that if we previously crashed due to a PANIC on fsync(), we'll be
 * rewriting all changes again during recovery.
 *
 * Note we assume we're chdir'd into PGDATA to begin with.
 */

	walkdir(".", datadir_fsync_fname, false, LOG);
	if (xlog_is_symlink)
		walkdir("pg_wal", datadir_fsync_fname, false, LOG);
	walkdir("pg_tblspc", datadir_fsync_fname, true, LOG);

	StartupPID = StartupDataBase();
	Assert(StartupPID != 0);
	StartupStatus = STARTUP_RUNNING;
	pmState = PM_STARTUP;  //状态机改变状态

	/* Some workers may be scheduled to start now */
	maybe_start_bgworkers();

	status = ServerLoop();

	/*
	 * ServerLoop probably shouldn't ever return, but if it does, close down.
	 */
	ExitPostmaster(status != STATUS_OK);

	abort();					/* not reached */
}

/*
 * We use a simple state machine to control startup, shutdown, and
 * crash recovery (which is rather like shutdown followed by startup).
 *
 * After doing all the postmaster initialization work, we enter PM_STARTUP
 * state and the startup process is launched. The startup process begins by
 * reading the control file and other preliminary initialization steps.
 * In a normal startup, or after crash recovery, the startup process exits
 * with exit code 0 and we switch to PM_RUN state.  

		if (pid == StartupPID)
		{
			StartupPID = 0;
...

			/*
			 * Startup succeeded, commence normal operations
			 */
			StartupStatus = STARTUP_NOT_RUNNING; //由STARTUP_RUNNING转成STARTUP_NOT_RUNNING
			FatalError = false; //上面一堆if未命中后，才不是fatal的
			AbortStartTime = 0;
			ReachedNormalRunning = true;
			pmState = PM_RUN; //状态机由PM_STARTUP转成PM_RUN
			connsAllowed = ALLOW_ALL_CONNS;

			/*
			 * Crank up the background tasks, if we didn't do that already
			 * when we entered consistent recovery state.  It doesn't matter
			 * if this fails, we'll just try again later.
			 */
            //以下都在启动核心子进程
			if (CheckpointerPID == 0)
				CheckpointerPID = StartCheckpointer();
			if (BgWriterPID == 0)
				BgWriterPID = StartBackgroundWriter();
			if (WalWriterPID == 0)
				WalWriterPID = StartWalWriter();

			/*
			 * Likewise, start other special children as needed.  In a restart
			 * situation, some of them may be alive already.
			 */
            //以下都在启动非核心子进程
			if (!IsBinaryUpgrade && AutoVacuumingActive() && AutoVacPID == 0)
				AutoVacPID = StartAutoVacLauncher();
			if (PgArchStartupAllowed() && PgArchPID == 0)
				PgArchPID = pgarch_start();
			if (PgStatPID == 0)
				PgStatPID = pgstat_start();

			/* workers may be scheduled to start now */
			maybe_start_bgworkers();
		    //此时才是正式的可接受连接状态
			/* at this point we are really open for business */
			ereport(LOG,
					(errmsg("database system is ready to accept connections")));

			/* Report status */
			AddToDataDirLockFile(LOCK_FILE_LINE_PM_STATUS, PM_STATUS_READY);
#ifdef USE_SYSTEMD
			sd_notify(0, "READY=1");
#endif

			continue;
		}

	if (pid < 0)
	{
		/* in parent, fork failed */
		int			save_errno = errno;

		errno = save_errno;
		switch (type)
		{
			case StartupProcess:
				ereport(LOG,
						(errmsg("could not fork startup process: %m")));
				break;
			case BgWriterProcess:
				ereport(LOG,
						(errmsg("could not fork background writer process: %m")));
				break;
			case CheckpointerProcess:
				ereport(LOG,
						(errmsg("could not fork checkpointer process: %m")));
				break;
			case WalWriterProcess:
				ereport(LOG,
						(errmsg("could not fork WAL writer process: %m")));
				break;
			case WalReceiverProcess:
				ereport(LOG,
						(errmsg("could not fork WAL receiver process: %m")));
				break;
			default:
				ereport(LOG,
						(errmsg("could not fork process: %m")));
				break;
		}

		/*
		 * fork failure is fatal during startup, but there's no need to choke
		 * immediately if starting other child types fails.
		 */
		if (type == StartupProcess)
			ExitPostmaster(1);
		return 0;
	}

static const Size max_changes_in_memory = 4096;

/*
 * Check whether the transaction tx should spill its data to disk.
 */
static void
ReorderBufferCheckSerializeTXN(ReorderBuffer *rb, ReorderBufferTXN *txn)
{
	/*
	 * TODO: improve accounting so we cheaply can take subtransactions into
	 * account here.
	 */
	if (txn->nentries_mem >= max_changes_in_memory)
	{
		ReorderBufferSerializeTXN(rb, txn);
		Assert(txn->nentries_mem == 0);
	}
}

static void
ReorderBufferCheckMemoryLimit(ReorderBuffer *rb)
{
...
	while (rb->size >= logical_decoding_work_mem * 1024L)
	{
		/*
		 * Pick the largest transaction (or subtransaction) and evict it from
		 * memory by serializing it to disk.
		 */
		txn = ReorderBufferLargestTXN(rb);

		ReorderBufferSerializeTXN(rb, txn);
...
}

static void
ReorderBufferCheckMemoryLimit(ReorderBuffer *rb)
{
...
	while (rb->size >= logical_decoding_work_mem * 1024L)
	{
		/*
		 * Pick the largest transaction (or subtransaction) and evict it from
		 * memory by streaming, if possible.  Otherwise, spill to disk.
		 */
		if (ReorderBufferCanStartStreaming(rb) &&
			(txn = ReorderBufferLargestTopTXN(rb)) != NULL)
		{...
			ReorderBufferStreamTXN(rb, txn);
		}
		else
		{...
			ReorderBufferSerializeTXN(rb, txn);
		}
...
}

/* Returns true, if the streaming can be started now, false, otherwise. */
static inline bool
ReorderBufferCanStartStreaming(ReorderBuffer *rb)
{
	LogicalDecodingContext *ctx = rb->private_data;
	SnapBuild  *builder = ctx->snapshot_builder;

	/* We can't start streaming unless a consistent state is reached. */
	if (SnapBuildCurrentState(builder) < SNAPBUILD_CONSISTENT)
		return false;

	/*
	 * We can't start streaming immediately even if the streaming is enabled
	 * because we previously decoded this transaction and now just are
	 * restarting.
	 */
	if (ReorderBufferCanStream(rb) &&
		!SnapBuildXactNeedsSkip(builder, ctx->reader->EndRecPtr))
		return true;

	return false;
}

	/*
	 * Found a point after SNAPBUILD_FULL_SNAPSHOT where all transactions that
	 * were running at that point finished. Till we reach that we hold off
	 * calling any commit callbacks.
	 */
	SNAPBUILD_CONSISTENT = 2

/*
 * Remove any leftover serialized reorder buffers from a slot directory after a
 * prior crash or decoding session exit.
 */
static void
ReorderBufferCleanupSerializedTXNs(const char *slotname)
{
	DIR		   *spill_dir;
	struct dirent *spill_de;
	struct stat statbuf;
	char		path[MAXPGPATH * 2 + 12];

	sprintf(path, "pg_replslot/%s", slotname);

	/* we're only handling directories here, skip if it's not ours */
	if (lstat(path, &statbuf) == 0 && !S_ISDIR(statbuf.st_mode))
		return;

	spill_dir = AllocateDir(path);
	while ((spill_de = ReadDirExtended(spill_dir, path, INFO)) != NULL)
	{
		/* only look at names that can be ours */
        //只对比前3个字符
		if (strncmp(spill_de->d_name, "xid", 3) == 0)
		{
			snprintf(path, sizeof(path),
					 "pg_replslot/%s/%s", slotname,
					 spill_de->d_name);

			if (unlink(path) != 0)
				ereport(ERROR,
						(errcode_for_file_access(),mkdir 
						 errmsg("could not remove file \"%s\" during removal of pg_replslot/%s/xid*: %m",
								path, slotname)));
		}
	}
	FreeDir(spill_dir);
}

void
StartupReorderBuffer(void)
{
	DIR		   *logical_dir;
	struct dirent *logical_de;

	logical_dir = AllocateDir("pg_replslot");
	while ((logical_de = ReadDir(logical_dir, "pg_replslot")) != NULL)
	{	//排除.和..
		if (strcmp(logical_de->d_name, ".") == 0 ||
			strcmp(logical_de->d_name, "..") == 0)
			continue;
		//验证slotname是否规范
		/* if it cannot be a slot, skip the directory */
		if (!ReplicationSlotValidateName(logical_de->d_name, DEBUG2))
			continue;

		/*
		 * ok, has to be a surviving logical slot, iterate and delete
		 * everything starting with xid-*
		 */
		ReorderBufferCleanupSerializedTXNs(logical_de->d_name);
	}
	FreeDir(logical_dir);
}

/*
 * Permanently drop the replication slot which will be released by the point
 * this function returns.
 */
static void
ReplicationSlotDropPtr(ReplicationSlot *slot)
{
	char		path[MAXPGPATH];
	char		tmppath[MAXPGPATH];

	/*
	 * If some other backend ran this code concurrently with us, we might try
	 * to delete a slot with a certain name while someone else was trying to
	 * create a slot with the same name.
	 */
	LWLockAcquire(ReplicationSlotAllocationLock, LW_EXCLUSIVE);

	/* Generate pathnames. */
	sprintf(path, "pg_replslot/%s", NameStr(slot->data.name));
	sprintf(tmppath, "pg_replslot/%s.tmp", NameStr(slot->data.name));

	/*
	 * Rename the slot directory on disk, so that we'll no longer recognize
	 * this as a valid slot.  Note that if this fails, we've got to mark the
	 * slot inactive before bailing out.  If we're dropping an ephemeral or a
	 * temporary slot, we better never fail hard as the caller won't expect
	 * the slot to survive and this might get called during error handling.
	 */
	if (rename(path, tmppath) == 0) //rename文件
	{
		/*
		 * We need to fsync() the directory we just renamed and its parent to
		 * make sure that our changes are on disk in a crash-safe fashion.  If
		 * fsync() fails, we can't be sure whether the changes are on disk or
		 * not.  For now, we handle that by panicking;
		 * StartupReplicationSlots() will try to straighten it out after
		 * restart.
		 */
        //fsync持久化
		START_CRIT_SECTION();
		fsync_fname(tmppath, true);
		fsync_fname("pg_replslot", true);
		END_CRIT_SECTION();
	}
...

	/*
	 * If removing the directory fails, the worst thing that will happen is
	 * that the user won't be able to create a new slot with the same name
	 * until the next server restart.  We warn about it, but that's all.
	 */
 
	if (!rmtree(tmppath, true))
		ereport(WARNING,
				(errmsg("could not remove directory \"%s\"", tmppath)));
	/*
	 * We release this at the very end, so that nobody starts trying to create
	 * a slot while we're still cleaning up the detritus of the old one.
	 */
	LWLockRelease(ReplicationSlotAllocationLock);
}

bool
ReplicationSlotValidateName(const char *name, int elevel)
{
...

	for (cp = name; *cp; cp++)
	{   //关键规则在这里
		if (!((*cp >= 'a' && *cp <= 'z')
			  || (*cp >= '0' && *cp <= '9')
			  || (*cp == '_')))
		{
			ereport(elevel,
					(errcode(ERRCODE_INVALID_NAME),
					 errmsg("replication slot name \"%s\" contains invalid character",
							name),
					 errhint("Replication slot names may only contain lower case letters, numbers, and the underscore character.")));
			return false;
		}
	}
	return true;
}

cd pg_replslot
mv slotname slotname.bak 
mkdir slotname
cp slotname.bak/state slotname/