本節簡單介紹了PostgreSQL的備份工具pg_basebackup原始碼中實際執行備份邏輯的BaseBackup中對WAL資料進行備份的實現函式StartLogStreamer.

### 一、資料結構
**logstreamer_param**
WAL data streamer引數.
```

typedef struct
{
     ////後臺連線
    PGconn     *bgconn;
    //開始位置
    XLogRecPtr  startptr;
    //目錄或者tar檔案,依賴於使用的模式
    char        xlog[MAXPGPATH];    /* directory or tarfile depending on mode */
    //系統識別符號
    char       *sysidentifier;
    //時間線
    int         timeline;
} logstreamer_param;

```

**StreamCtl**
接收xlog流資料時的全域性引數
```
/*
 * Global parameters when receiving xlog stream. For details about the individual fields,
 * see the function comment for ReceiveXlogStream().
 * 接收xlog流資料時的全域性引數.
 * 每個域欄位的詳細解釋,參見ReceiveXlogStream()函式註釋.
 */
typedef struct StreamCtl
{
    //streaming的開始位置
    XLogRecPtr  startpos;       /* Start position for streaming */
    //時間線
    TimeLineID  timeline;       /* Timeline to stream data from */
    //系統識別符號
    char       *sysidentifier;  /* Validate this system identifier and
                                 * timeline */
    //standby超時資訊
    int         standby_message_timeout;    /* Send status messages this often */
    //是否同步(寫入時是否馬上Flush WAL data)
    bool        synchronous;    /* Flush immediately WAL data on write */
    //在已歸檔的資料中標記segment為已完成
    bool        mark_done;      /* Mark segment as done in generated archive */
    //重新整理到磁碟上以確保資料的一致性狀態(是否已重新整理到磁碟上)
    bool        do_sync;        /* Flush to disk to ensure consistent state of
                                 * data */
    //在返回T時停止streaming
    stream_stop_callback stream_stop;   /* Stop streaming when returns true */
    //如有效,監測該socket中的輸入並檢查stream_stop()的返回
    pgsocket    stop_socket;    /* if valid, watch for input on this socket
                                 * and check stream_stop() when there is any */
    //如何寫WAL
    WalWriteMethod *walmethod;  /* How to write the WAL */
    //附加到部分接受檔案的字尾
    char       *partial_suffix; /* Suffix appended to partially received files */
    //使用的replication slot,如無則為NULL
    char       *replication_slot;   /* Replication slot to use, or NULL */
} StreamCtl;

```

### 二、原始碼解讀
**StartLogStreamer**
StartLogStreamer用於在備份時初始化後臺程式用於接收WAL.接收程式將建立自己的資料庫連線以並行的方式對檔案進行streaming複製.
```
/*
 * Initiate background process for receiving xlog during the backup.
 * The background stream will use its own database connection so we can
 * stream the logfile in parallel with the backups.
 * 在備份時初始化後臺程式用於接收WAL.
 * 後臺stream程式將用自己的資料庫連線以使以並行的方式stream檔案.
 */
static void
StartLogStreamer(char *startpos, uint32 timeline, char *sysidentifier)
{
    //引數
    logstreamer_param *param;
    uint32      hi,
                lo;//高位/低位
    char        statusdir[MAXPGPATH];

param = pg_malloc0(sizeof(logstreamer_param));
    param->timeline = timeline;
    param->sysidentifier = sysidentifier;

/* Convert the starting position */
    //轉換開始位置(高低位轉換)
    if (sscanf(startpos, "%X/%X", &hi, &lo) != 2)
    {
        fprintf(stderr,
                _("%s: could not parse write-ahead log location \"%s\"\n"),
                progname, startpos);
        exit(1);
    }
    //開始位置,轉換為64bit的地址
    param->startptr = ((uint64) hi) << 32 | lo;
    /* Round off to even segment position */
    //按segment取整
    param->startptr -= XLogSegmentOffset(param->startptr, WalSegSz);

#ifndef WIN32
    //WIN32使用的程式碼
    /* Create our background pipe */
    if (pipe(bgpipe) < 0)
    {
        fprintf(stderr,
                _("%s: could not create pipe for background process: %s\n"),
                progname, strerror(errno));
        exit(1);
    }
#endif

/* Get a second connection */
    //獲取第二個連線
    param->bgconn = GetConnection();
    if (!param->bgconn)
        /* Error message already written in GetConnection() */
        exit(1);

/* In post-10 cluster, pg_xlog has been renamed to pg_wal */
    //在PG 10,pg_xlog已命名為pg_wal
    snprintf(param->xlog, sizeof(param->xlog), "%s/%s",
             basedir,
             PQserverVersion(conn) < MINIMUM_VERSION_FOR_PG_WAL ?
             "pg_xlog" : "pg_wal");

/* Temporary replication slots are only supported in 10 and newer */
    //臨時複製slots只在PG10+支援
    if (PQserverVersion(conn) < MINIMUM_VERSION_FOR_TEMP_SLOTS)
        temp_replication_slot = false;

/*
     * Create replication slot if requested
     * 如要求,則建立複製slot
     */
    //static char *replication_slot = NULL;
    //static bool temp_replication_slot = true;
    if (temp_replication_slot && !replication_slot)
        //建立replication slot
        replication_slot = psprintf("pg_basebackup_%d", (int) PQbackendPID(param->bgconn));
    if (temp_replication_slot || create_slot)
    {
        //建立replication slot
        if (!CreateReplicationSlot(param->bgconn, replication_slot, NULL,
                                   temp_replication_slot, true, true, false))
            exit(1);

if (verbose)
        {
            //顯示診斷資訊
            if (temp_replication_slot)
                fprintf(stderr, _("%s: created temporary replication slot \"%s\"\n"),
                        progname, replication_slot);
            else
                fprintf(stderr, _("%s: created replication slot \"%s\"\n"),
                        progname, replication_slot);
        }
    }

if (format == 'p')
    {
        /*
         * Create pg_wal/archive_status or pg_xlog/archive_status (and thus
         * pg_wal or pg_xlog) depending on the target server so we can write
         * to basedir/pg_wal or basedir/pg_xlog as the directory entry in the
         * tar file may arrive later.
         * 基於目標伺服器建立pg_wal/archive_status或pg_xlog/archive_status,
         * 這樣可以寫入到basedir/pg_wal 貨 basedir/pg_xlog,可作為後續訪問的tar檔案目錄條目
         */
        snprintf(statusdir, sizeof(statusdir), "%s/%s/archive_status",
                 basedir,
                 PQserverVersion(conn) < MINIMUM_VERSION_FOR_PG_WAL ?
                 "pg_xlog" : "pg_wal");

if (pg_mkdir_p(statusdir, pg_dir_create_mode) != 0 && errno != EEXIST)
        {
            fprintf(stderr,
                    _("%s: could not create directory \"%s\": %s\n"),
                    progname, statusdir, strerror(errno));
            exit(1);
        }
    }

/*
     * Start a child process and tell it to start streaming. On Unix, this is
     * a fork(). On Windows, we create a thread.
     * 啟動子程式開始streaming.
     * 在UNIX平臺,是一個fork程式,在Windows平臺,建立執行緒.
     */
#ifndef WIN32
    //UNIX:fork程式
    bgchild = fork();
    if (bgchild == 0)
    {
        //這是子程式,返回0
        /* in child process */
        //啟動新程式
        exit(LogStreamerMain(param));
    }
    else if (bgchild < 0)
    {
        fprintf(stderr, _("%s: could not create background process: %s\n"),
                progname, strerror(errno));
        exit(1);
    }

/*
     * Else we are in the parent process and all is well.
     * 在父程式中,返回的bgchild是子程式PID.
     */
    atexit(kill_bgchild_atexit);
#else                           /* WIN32 */
    //WIN32:建立執行緒
    bgchild = _beginthreadex(NULL, 0, (void *) LogStreamerMain, param, 0, NULL);
    if (bgchild == 0)
    {
        fprintf(stderr, _("%s: could not create background thread: %s\n"),
                progname, strerror(errno));
        exit(1);
    }
#endif
}

```

**LogStreamerMain**
WAL流複製主函式,用於fork後的子程式呼叫
```
static int
LogStreamerMain(logstreamer_param *param)
{
    StreamCtl   stream;//接收xlog流資料時的全域性引數

in_log_streamer = true;

//初始化StreamCtl結構體
    MemSet(&stream, 0, sizeof(stream));
    stream.startpos = param->startptr;
    stream.timeline = param->timeline;
    stream.sysidentifier = param->sysidentifier;
    stream.stream_stop = reached_end_position;
#ifndef WIN32
    stream.stop_socket = bgpipe[0];
#else
    stream.stop_socket = PGINVALID_SOCKET;
#endif
    stream.standby_message_timeout = standby_message_timeout;
    stream.synchronous = false;
    stream.do_sync = do_sync;
    stream.mark_done = true;
    stream.partial_suffix = NULL;
    stream.replication_slot = replication_slot;

if (format == 'p')
        stream.walmethod = CreateWalDirectoryMethod(param->xlog, 0, do_sync);
    else
        stream.walmethod = CreateWalTarMethod(param->xlog, compresslevel, do_sync);
    //接收資料
    if (!ReceiveXlogStream(param->bgconn, &stream))

/*
         * Any errors will already have been reported in the function process,
         * but we need to tell the parent that we didn't shutdown in a nice
         * way.
         * 在函式執行過程中出現的錯誤已透過警告的方式發出,
         * 但仍需要告知父程式不能優雅的關閉本程式.
         */
        return 1;

if (!stream.walmethod->finish())
    {
        fprintf(stderr,
                _("%s: could not finish writing WAL files: %s\n"),
                progname, strerror(errno));
        return 1;
    }
    //結束連線
    PQfinish(param->bgconn);
    //普通檔案格式
    if (format == 'p')
        FreeWalDirectoryMethod();
    else
        FreeWalTarMethod();
    //是否記憶體
    pg_free(stream.walmethod);

return 0;
}
 
```

### 三、跟蹤分析
備份命令
```
pg_basebackup -h 192.168.26.25 -U replicator -p 5432 -D /data/backup -P -Xs -R -v

```
啟動gdb跟蹤
```
[xdb@localhost ~]$ gdb pg_basebackup
GNU gdb (GDB) Red Hat Enterprise Linux 7.6.1-110.el7
Copyright (C) 2013 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http:>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.  Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-redhat-linux-gnu".
For bug reporting instructions, please see:
<http:></http:>...
Reading symbols from /appdb/atlasdb/pg11.2/bin/pg_basebackup...done.
(gdb) b StartLogStreamer
Breakpoint 1 at 0x403e6b: file pg_basebackup.c, line 555.
(gdb) set args -h 192.168.26.25 -U replicator -p 5432 -D /data/backup -P -Xs -R -v
(gdb) r
Starting program: /appdb/xdb/pg11.2/bin/pg_basebackup -h 192.168.26.25 -U replicator -p 5432 -D /data/backup -P -Xs -R -v
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
Password: 
pg_basebackup: initiating base backup, waiting for checkpoint to complete
pg_basebackup: checkpoint completed
pg_basebackup: write-ahead log start point: 0/57000060 on timeline 16
pg_basebackup: starting background WAL receiver

Breakpoint 1, StartLogStreamer (startpos=0x7fffffffdf60 "0/57000060", timeline=16, 
    sysidentifier=0x61f1a0 "6666964067616600474") at pg_basebackup.c:555
555     param = pg_malloc0(sizeof(logstreamer_param));
(gdb) 
```
輸入引數
startpos=0x7fffffffdf60 "0/57000060", 
timeline=16, 
sysidentifier=0x61f1a0 "6666964067616600474"
構造引數
```
(gdb) n
556     param->timeline = timeline;
(gdb) 
557     param->sysidentifier = sysidentifier;
(gdb) 
560     if (sscanf(startpos, "%X/%X", &hi, &lo) != 2)
(gdb) 
567     param->startptr = ((uint64) hi) << 32 | lo;
(gdb) p hi
$1 = 0
(gdb) p lo
$2 = 1459617888
(gdb) n
569     param->startptr -= XLogSegmentOffset(param->startptr, WalSegSz);
(gdb) n
573     if (pipe(bgpipe) < 0)
(gdb) p *param
$3 = {bgconn = 0x0, startptr = 1459617792, xlog = '\000' <repeats times="">, 
  sysidentifier = 0x61f1a0 "6666964067616600474", timeline = 16}
(gdb) 
```
建立連線,建立replication slot
```
(gdb) n
583     param->bgconn = GetConnection();
(gdb) 
584     if (!param->bgconn)
(gdb) 
591              PQserverVersion(conn) < MINIMUM_VERSION_FOR_PG_WAL ?
(gdb) 
589     snprintf(param->xlog, sizeof(param->xlog), "%s/%s",
(gdb) 
595     if (PQserverVersion(conn) < MINIMUM_VERSION_FOR_TEMP_SLOTS)
(gdb) 
601     if (temp_replication_slot && !replication_slot)
(gdb) 
602         replication_slot = psprintf("pg_basebackup_%d", (int) PQbackendPID(param->bgconn));
(gdb) 
603     if (temp_replication_slot || create_slot)
(gdb) 
605         if (!CreateReplicationSlot(param->bgconn, replication_slot, NULL,
(gdb) 
609         if (verbose)
(gdb) 
611             if (temp_replication_slot)
(gdb) 
612                 fprintf(stderr, _("%s: created temporary replication slot \"%s\"\n"),
(gdb) 
pg_basebackup: created temporary replication slot "pg_basebackup_59378"
620     if (format == 'p')
(gdb) 
(gdb) n
630                  PQserverVersion(conn) < MINIMUM_VERSION_FOR_PG_WAL ?
(gdb) 
628         snprintf(statusdir, sizeof(statusdir), "%s/%s/archive_status",
```
建立備份目錄
```
(gdb) 
633         if (pg_mkdir_p(statusdir, pg_dir_create_mode) != 0 && errno != EEXIST)
(gdb) p *param
$4 = {bgconn = 0x62a280, startptr = 1459617792, xlog = "/data/backup/pg_wal", '\000' <repeats times="">, 
  sysidentifier = 0x61f1a0 "6666964067616600474", timeline = 16}
(gdb) n
647     bgchild = fork();
(gdb) 
#############
[xdb@localhost backup]$ ls
pg_wal
```
fork程式,父程式返回子程式的PID
```
(gdb) n
647     bgchild = fork();
(gdb) n
Detaching after fork from child process 43001.
648     if (bgchild == 0)
(gdb) p bgchild
$5 = 43001
(gdb) 
```
子程式(PID=43001)
```
[xdb@localhost backup]$ ps -ef|grep 43001
xdb      43001 42820  1 11:54 pts/1    00:00:01 /appdb/xdb/pg11.2/bin/pg_basebackup -h 192.168.26.25 -U replicator -p 5432 -D /data/backup -P -Xs -R -v
[xdb@localhost backup]$ ps -ef|grep 192.168.26.25
xdb      42820 42756  0 11:48 pts/1    00:00:00 /appdb/xdb/pg11.2/bin/pg_basebackup -h 192.168.26.25 -U replicator -p 5432 -D /data/backup -P -Xs -R -v
xdb      43001 42820  0 11:54 pts/1    00:00:01 /appdb/xdb/pg11.2/bin/pg_basebackup -h 192.168.26.25 -U replicator -p 5432 -D /data/backup -P -Xs -R -v
```
完成呼叫
```
(gdb) n
653     else if (bgchild < 0)
(gdb) 
672 }
(gdb) 
BaseBackup () at pg_basebackup.c:1937
1937        for (i = 0; i < PQntuples(res); i++)
(gdb) 
```
pg_wal目錄中的資料
```
[xdb@localhost backup]$ ls -l ./pg_wal/
total 16388
-rw-------. 1 xdb xdb 16777216 Mar 18 11:54 000000100000000000000057
-rw-------. 1 xdb xdb      217 Mar 18 11:54 00000010.history
drwx------. 2 xdb xdb       35 Mar 18 11:54 archive_status
[xdb@localhost backup]$ 
```
DONE!

### 四、參考資料
[PG Source Code](https://doxygen.postgresql.org)</repeats></repeats></http:>

PostgreSQL 原始碼解讀（151）- PG Tools#3（StartLogStreamer）

相關文章