Redis集群分析（10）

1、 AOF重写源码分析

在（9）中分析了aof的重写机制和部分源码。这里我们继续分析，在子线程完成了对AOF文件的重写后，他实际写入的文件是temp-rewriteaof-bg-%d.aof，生成的文件只有在重写开始是的数据。按照之前分析的机制来说，还需要将重写buf中的数据写到文件中，并将文件改为AOF文件的名称。这部分操作实际是在父进程中完成的。

父进程在fork结束后对子进程的监控方式与RDB中的方式是一样的，都在serverCron方法中，其代码如下：

与RDB不同的是它执行的方法为backgroundRewriteDoneHandler。这个方法在aof.c中实现，其内容如下：

/* A background append only file rewriting (BGREWRITEAOF) terminated its work.
 * Handle this. */
void backgroundRewriteDoneHandler(int exitcode, int bysignal) {
    if (!bysignal && exitcode == 0) {
        int newfd, oldfd;
        char tmpfile[256];
        long long now = ustime();
        mstime_t latency;

        serverLog(LL_NOTICE,
            "Background AOF rewrite terminated with success");

        /* Flush the differences accumulated by the parent to the
         * rewritten AOF. */
        latencyStartMonitor(latency);
        snprintf(tmpfile,256,"temp-rewriteaof-bg-%d.aof",
            (int)server.aof_child_pid);
        newfd = open(tmpfile,O_WRONLY|O_APPEND);
        if (newfd == -1) {
            serverLog(LL_WARNING,
                "Unable to open the temporary AOF produced by the child: %s", strerror(errno));
            goto cleanup;
        }

        if (aofRewriteBufferWrite(newfd) == -1) {
            serverLog(LL_WARNING,
                "Error trying to flush the parent diff to the rewritten AOF: %s", strerror(errno));
            close(newfd);
            goto cleanup;
        }
        latencyEndMonitor(latency);
        latencyAddSampleIfNeeded("aof-rewrite-diff-write",latency);

        serverLog(LL_NOTICE,
            "Residual parent diff successfully flushed to the rewritten AOF (%.2f MB)", (double) aofRewriteBufferSize() / (1024*1024));

        /* The only remaining thing to do is to rename the temporary file to
         * the configured file and switch the file descriptor used to do AOF
         * writes. We don't want close(2) or rename(2) calls to block the
         * server on old file deletion.
         *
         * There are two possible scenarios:
         *
         * 1) AOF is DISABLED and this was a one time rewrite. The temporary
         * file will be renamed to the configured file. When this file already
         * exists, it will be unlinked, which may block the server.
         *
         * 2) AOF is ENABLED and the rewritten AOF will immediately start
         * receiving writes. After the temporary file is renamed to the
         * configured file, the original AOF file descriptor will be closed.
         * Since this will be the last reference to that file, closing it
         * causes the underlying file to be unlinked, which may block the
         * server.
         *
         * To mitigate the blocking effect of the unlink operation (either
         * caused by rename(2) in scenario 1, or by close(2) in scenario 2), we
         * use a background thread to take care of this. First, we
         * make scenario 1 identical to scenario 2 by opening the target file
         * when it exists. The unlink operation after the rename(2) will then
         * be executed upon calling close(2) for its descriptor. Everything to
         * guarantee atomicity for this switch has already happened by then, so
         * we don't care what the outcome or duration of that close operation
         * is, as long as the file descriptor is released again. */
        if (server.aof_fd == -1) {
            /* AOF disabled */

            /* Don't care if this fails: oldfd will be -1 and we handle that.
             * One notable case of -1 return is if the old file does
             * not exist. */
            oldfd = open(server.aof_filename,O_RDONLY|O_NONBLOCK);
        } else {
            /* AOF enabled */
            oldfd = -1; /* We'll set this to the current AOF filedes later. */
        }

        /* Rename the temporary file. This will not unlink the target file if
         * it exists, because we reference it with "oldfd". */
        latencyStartMonitor(latency);
        if (rename(tmpfile,server.aof_filename) == -1) {
            serverLog(LL_WARNING,
                "Error trying to rename the temporary AOF file %s into %s: %s",
                tmpfile,
                server.aof_filename,
                strerror(errno));
            close(newfd);
            if (oldfd != -1) close(oldfd);
            goto cleanup;
        }
        latencyEndMonitor(latency);
        latencyAddSampleIfNeeded("aof-rename",latency);

        if (server.aof_fd == -1) {
            /* AOF disabled, we don't need to set the AOF file descriptor
             * to this new file, so we can close it. */
            close(newfd);
        } else {
            /* AOF enabled, replace the old fd with the new one. */
            oldfd = server.aof_fd;
            server.aof_fd = newfd;
            if (server.aof_fsync == AOF_FSYNC_ALWAYS)
                redis_fsync(newfd);
            else if (server.aof_fsync == AOF_FSYNC_EVERYSEC)
                aof_background_fsync(newfd);
            server.aof_selected_db = -1; /* Make sure SELECT is re-issued */
            aofUpdateCurrentSize();
            server.aof_rewrite_base_size = server.aof_current_size;
            server.aof_fsync_offset = server.aof_current_size;

            /* Clear regular AOF buffer since its contents was just written to
             * the new AOF from the background rewrite buffer. */
            sdsfree(server.aof_buf);
            server.aof_buf = sdsempty();
        }

        server.aof_lastbgrewrite_status = C_OK;

        serverLog(LL_NOTICE, "Background AOF rewrite finished successfully");
        /* Change state from WAIT_REWRITE to ON if needed */
        if (server.aof_state == AOF_WAIT_REWRITE)
            server.aof_state = AOF_ON;

        /* Asynchronously close the overwritten AOF. */
        if (oldfd != -1) bioCreateBackgroundJob(BIO_CLOSE_FILE,(void*)(long)oldfd,NULL,NULL);

        serverLog(LL_VERBOSE,
            "Background AOF rewrite signal handler took %lldus", ustime()-now);
    } else if (!bysignal && exitcode != 0) {
        server.aof_lastbgrewrite_status = C_ERR;

        serverLog(LL_WARNING,
            "Background AOF rewrite terminated with error");
    } else {
        /* SIGUSR1 is whitelisted, so we have a way to kill a child without
         * tirggering an error condition. */
        if (bysignal != SIGUSR1)
            server.aof_lastbgrewrite_status = C_ERR;

        serverLog(LL_WARNING,
            "Background AOF rewrite terminated by signal %d", bysignal);
    }

cleanup:
    aofClosePipes();
    aofRewriteBufferReset();
    aofRemoveTempFile(server.aof_child_pid);
    server.aof_child_pid = -1;
    server.aof_rewrite_time_last = time(NULL)-server.aof_rewrite_time_start;
    server.aof_rewrite_time_start = -1;
    /* Schedule a new rewrite if we are waiting for it to switch the AOF ON. */
    if (server.aof_state == AOF_WAIT_REWRITE)
        server.aof_rewrite_scheduled = 1;
}

这个方法看起来很长，但只有几个关键点。首先是第17行，这里通过pid拿到了子进程生成的文件的文件名，然后是第26行，这里调用aofRewriteBufferWrite方法将重写buf中的数据写到临时文件中。aofRewriteBufferWrite方法的实现方式如下：

/* Write the buffer (possibly composed of multiple blocks) into the specified
 * fd. If a short write or any other error happens -1 is returned,
 * otherwise the number of bytes written is returned. */
ssize_t aofRewriteBufferWrite(int fd) {
    listNode *ln;
    listIter li;
    ssize_t count = 0;

    listRewind(server.aof_rewrite_buf_blocks,&li);
    while((ln = listNext(&li))) {
        aofrwblock *block = listNodeValue(ln);
        ssize_t nwritten;

        if (block->used) {
            nwritten = write(fd,block->buf,block->used);
            if (nwritten != (ssize_t)block->used) {
                if (nwritten == 0) errno = EIO;
                return -1;
            }
            count += nwritten;
        }
    }
    return count;
}

这个方法实现了将重写buf中的数据写到文件中的操作，其中重写buf为server.aof_rewrite_buf_blocks，需要说明的是这个buf并不是一块固定大小的内存块，而是一个链表。这是因为无法确定子进程重写文件到底需要多少时间，也无法确定这段时间内会产生多少数据，所以使用一个链表来存储较为合适。

这个链表的每个节点存储的数据结构为aofrwblock，这个结构定义在aof.c文件中，其内容如下：

#define AOF_RW_BUF_BLOCK_SIZE (1024*1024*10)    /* 10 MB per block */

typedef struct aofrwblock {
    unsigned long used, free;
    char buf[AOF_RW_BUF_BLOCK_SIZE];
} aofrwblock;

这里可以看到每个aofrwblock中是有一个10MB的内存块，在写数据的时候会先拿到链表的最后一个block，然后向block中的buf写数据，当buf中写满了后，再创建一个新的block添加到链表的尾部。

在文档（8）中，我们提到了redis在将命令写到AOF文件的同时，会判断是否在进行AOF重写，若是则会将命令写到重写buf中。这里将命令写到重写buf中的方法为aofRewriteBufferAppend，这个方法内容如下：

/* Append data to the AOF rewrite buffer, allocating new blocks if needed. */
void aofRewriteBufferAppend(unsigned char *s, unsigned long len) {
    listNode *ln = listLast(server.aof_rewrite_buf_blocks);
    aofrwblock *block = ln ? ln->value : NULL;

    while(len) {
        /* If we already got at least an allocated block, try appending
         * at least some piece into it. */
        if (block) {
            unsigned long thislen = (block->free < len) ? block->free : len;
            if (thislen) {  /* The current block is not already full. */
                memcpy(block->buf+block->used, s, thislen);
                block->used += thislen;
                block->free -= thislen;
                s += thislen;
                len -= thislen;
            }
        }

        if (len) { /* First block to allocate, or need another block. */
            int numblocks;

            block = zmalloc(sizeof(*block));
            block->free = AOF_RW_BUF_BLOCK_SIZE;
            block->used = 0;
            listAddNodeTail(server.aof_rewrite_buf_blocks,block);

            /* Log every time we cross more 10 or 100 blocks, respectively
             * as a notice or warning. */
            numblocks = listLength(server.aof_rewrite_buf_blocks);
            if (((numblocks+1) % 10) == 0) {
                int level = ((numblocks+1) % 100) == 0 ? LL_WARNING :
                                                         LL_NOTICE;
                serverLog(level,"Background AOF buffer size: %lu MB",
                    aofRewriteBufferSize()/(1024*1024));
            }
        }
    }

    /* Install a file event to send data to the rewrite child if there is
     * not one already. */
    if (aeGetFileEvents(server.el,server.aof_pipe_write_data_to_child) == 0) {
        aeCreateFileEvent(server.el, server.aof_pipe_write_data_to_child,
            AE_WRITABLE, aofChildWriteDiffData, NULL);
    }
}

这个方法就和上文分析的一样，首先是第3行，拿到链表的最后一个节点。然后是一个while循环，循环结束条件是将所有的数据写入到重写buf中。然后是第9行，直接向拿到block中写数据，然后是第20行，如果block写满了但数据还没写完，就再创建一个block，并添加到链表的尾部。

分析完AOF的重写buf后，再看上文的aofRewriteBufferWrite方法就很简单了，就是将链表中的数据全部遍历出来并写到文件中。

在backgroundRewriteDoneHandler方法中，还有最后一个关键点在第80行，这里会将临时文件重命名为AOF文件，这个文件名由参数server.aof_filename来决定，其默认值为“appendonly.aof”。