最近準(zhǔn)備重寫以前的監(jiān)控日志文件變動的項目，以前監(jiān)控日志文件變動使用的技術(shù)基礎(chǔ)是jdk1.7新出的WatchService移袍，但是使用了接近一年以后發(fā)現(xiàn)了幾個問題：

1.無法指定專門文件監(jiān)聽爷绘，只能對整個目錄的所有文件進(jìn)行監(jiān)聽书劝，然后自己過濾
2.當(dāng)修改了目錄或者文件的任何metadata后，就不會再傳入任何變動事件給監(jiān)聽者了
3.無法很好的控制監(jiān)控頻率

對于以上的三個問題揉阎，使用WatchService技術(shù)無法很好的解決他們庄撮，所以到處找別的解決方案背捌。
因為我使用的是flume, 版本是1.6毙籽，當(dāng)我看到flume1.7使用的TaildirSource的時候，忍不住眼前一亮毡庆，這就是我需要的解決方案坑赡。但是使用之前我要判斷下是否有坑,于是去讀了下他的源代碼烙如。

TaildirSource的代碼很簡單，主要就是五個類:
ReliableTaildirEventReader類
TaildirMatcher類
TaildirSource類
TaildirSourceConfigurationConstants類
TailFile類

這五個類之間的關(guān)系描述如下:
TailSource不用說,Flume的啟動點(diǎn)
在其process方法中
TailSource調(diào)用ReliableTaildirEventReader的updateTailFiles方法去獲取所有需要關(guān)注的文件毅否，并讀取解析
而為了只關(guān)注需要關(guān)注的文件亚铁，在ReliableTaildirEventReader的updateTailFiles方法中使用了TaildirMatcher去過濾出需要的文件
而TaildirMatcher和文件系統(tǒng)的目錄是一一對應(yīng)的，一個目錄可以抽象為一個TaildirMatcher螟加，而被關(guān)注的文件被抽象為TailFile對象
TaildirSourceConfigurationConstants類不用多說徘溢，一些配置常量而已

理解了類和代碼，我們就可以明白整套TaildirSource動態(tài)監(jiān)聽文件變化的技術(shù)基礎(chǔ)就是獲取文件的inode,建立inode和文件之間的一一對應(yīng)關(guān)系捆探，利用RandomAccessFile去讀取文件然爆，并將inode和讀取的位置以及文件位置保存成json文件進(jìn)行持久化，以便后續(xù)的繼續(xù)跟蹤黍图。
但是曾雕，注意，inode是linux文件的概念助被，而獲取inode是在ReliableTaildirEventReader的getInode方法里剖张，在這個方法里，我們將受到一萬點(diǎn)暴擊:(long) Files.getAttribute(file.toPath(), "unix:ino");
這段代碼無恥的排除了windows操作系統(tǒng)的存在揩环。

看明白了TaildirSource的代碼搔弄，發(fā)現(xiàn)TaildirSource是不支持Windows的。怎么辦呢丰滑？首先想想解決思路:
我們知道整體TaildirSource的思想是獲取一個文件的標(biāo)識(linux里inode可以作為文件的標(biāo)識使用肯污，當(dāng)系統(tǒng)讀取文件時，其實就是根據(jù)文件路徑轉(zhuǎn)換成對應(yīng)的inode值來做的操作)并記錄對應(yīng)的文件路徑吨枉，那么我們首先要確認(rèn)在Windows中是否是有類似于inode這種東西的存在蹦渣。這個回答解答了這個問題，windows中是有file id這種類似于inode的存在的貌亭。

那么繼續(xù),這個file id是否是有什么限制或者有什么特性呢?
參考這個回答 我們知道file id是跟文件系統(tǒng)有關(guān)的, 在FAT系統(tǒng)中柬唯，如果修改的名字長于舊名字，file id可能會發(fā)生改變圃庭，但是在NTFS系統(tǒng)中锄奢，在刪除之前file id都是穩(wěn)定的。

ok剧腻，方案可以確定了拘央，如果是windows系統(tǒng) 并且文件系統(tǒng)是ntfs，那么我們就使用file id去獲取文件书在，剩下的邏輯幾乎和linux是一模一樣灰伟。
如果是fat系統(tǒng)(在我們的工作環(huán)境中不可能出現(xiàn))，我們暫時不做支持儒旬。
(另外提一句栏账，在windows 2012中新增加了一個Refs 這個由于我們基本不用帖族，所以沒有做考慮，參考這個)
方案落地如下:
使用java的jna
Kernel32的代碼:

package com.creditease.ns.jna;/* Copyright (c) 2007, 2013 Timothy Wall, Markus Karg, All Rights Reserved
 *
 * The contents of this file is dual-licensed under 2 
 * alternative Open Source/Free licenses: LGPL 2.1 or later and 
 * Apache License 2.0. (starting with JNA version 4.0.0).
 * 
 * You can freely decide which license you want to apply to 
 * the project.
 * 
 * You may obtain a copy of the LGPL License at:
 * 
 * http://www.gnu.org/licenses/licenses.html
 * 
 * A copy is also included in the downloadable source code package
 * containing JNA, in file "LGPL2.1".
 * 
 * You may obtain a copy of the Apache License at:
 * 
 * http://www.apache.org/licenses/
 * 
 * A copy is also included in the downloadable source code package
 * containing JNA, in file "AL2.0".
 */

import com.sun.jna.Native;
import com.sun.jna.Structure;
import com.sun.jna.platform.win32.WinBase;
import com.sun.jna.platform.win32.WinNT;
import com.sun.jna.platform.win32.Wincon;
import com.sun.jna.win32.StdCallLibrary;
import com.sun.jna.win32.W32APIOptions;

import java.util.ArrayList;
import java.util.Arrays;
import java.util.List;

/**
 * Interface definitions for <code>kernel32.dll</code>. Includes additional
 * alternate mappings from {@link WinNT} which make use of NIO buffers,
 * {@link Wincon} for console API.
 */
public interface Kernel32 extends StdCallLibrary, WinNT, Wincon {

    /**
     * The instance.
     */
    Kernel32 INSTANCE = (Kernel32) Native.loadLibrary("kernel32", Kernel32.class, W32APIOptions.DEFAULT_OPTIONS);

    /**
     * The GetLastError function retrieves the calling thread's last-error code
     * value. The last-error code is maintained on a per-thread basis. Multiple
     * threads do not overwrite each other's last-error code.
     *
     * @return The return value is the calling thread's last-error code value.
     */
    int GetLastError();

    /**
     * The CreateFile function creates or opens a file, file stream, directory,
     * physical disk, volume, console buffer, tape drive, communications
     * resource, mailslot, or named pipe. The function returns a handle that can
     * be used to access an object.
     *
     * @param lpFileName            A pointer to a null-terminated string that specifies the name
     *                              of an object to create or open.
     * @param dwDesiredAccess       The access to the object, which can be read, write, or both.
     * @param dwShareMode           The sharing mode of an object, which can be read, write, both,
     *                              or none.
     * @param lpSecurityAttributes  A pointer to a SECURITY_ATTRIBUTES structure that determines
     *                              whether or not the returned handle can be inherited by child
     *                              processes. If lpSecurityAttributes is NULL, the handle cannot
     *                              be inherited.
     * @param dwCreationDisposition An action to take on files that exist and do not exist.
     * @param dwFlagsAndAttributes  The file attributes and flags.
     * @param hTemplateFile         Handle to a template file with the GENERIC_READ access right.
     *                              The template file supplies file attributes and extended
     *                              attributes for the file that is being created. This parameter
     *                              can be NULL.
     * @return If the function succeeds, the return value is an open handle to a
     * specified file. If a specified file exists before the function
     * call and dwCreationDisposition is CREATE_ALWAYS or OPEN_ALWAYS, a
     * call to GetLastError returns ERROR_ALREADY_EXISTS, even when the
     * function succeeds. If a file does not exist before the call,
     * GetLastError returns 0 (zero). If the function fails, the return
     * value is INVALID_HANDLE_VALUE. To get extended error information,
     * call GetLastError.
     */
    HANDLE CreateFile(String lpFileName, int dwDesiredAccess, int dwShareMode,
        WinBase.SECURITY_ATTRIBUTES lpSecurityAttributes,
        int dwCreationDisposition, int dwFlagsAndAttributes,
        HANDLE hTemplateFile);


    /**
     * Closes an open object handle.
     *
     * @param hObject Handle to an open object. This parameter can be a pseudo
     *                handle or INVALID_HANDLE_VALUE.
     * @return If the function succeeds, the return value is nonzero. If the
     * function fails, the return value is zero. To get extended error
     * information, call {@code GetLastError}.
     * @see <A >CloseHandle</A>
     */
    boolean CloseHandle(HANDLE hObject);


    class BY_HANDLE_FILE_INFORMATION extends Structure {
        public DWORD dwFileAttributes;
        public FILETIME ftCreationTime;
        public FILETIME ftLastAccessTime;
        public FILETIME ftLastWriteTime;
        public DWORD dwVolumeSerialNumber;
        public DWORD nFileSizeHigh;
        public DWORD nFileSizeLow;
        public DWORD nNumberOfLinks;
        public DWORD nFileIndexHigh;
        public DWORD nFileIndexLow;

        public static class ByReference extends BY_HANDLE_FILE_INFORMATION
            implements Structure.ByReference {

        }


        public static class ByValue extends BY_HANDLE_FILE_INFORMATION
            implements Structure.ByValue {

        }

        @Override
        protected List getFieldOrder() {
            List fields = new ArrayList();
            fields.addAll(Arrays.asList(new String[]{"dwFileAttributes",
                "ftCreationTime", "ftLastAccessTime", "ftLastWriteTime",
                "dwVolumeSerialNumber", "nFileSizeHigh", "nFileSizeLow",
                "nNumberOfLinks", "nFileIndexHigh", "nFileIndexLow"}));
            return fields;

        }
    }

    boolean GetFileInformationByHandle(HANDLE hFile,
        BY_HANDLE_FILE_INFORMATION lpFileInformation);
}

package com.creditease.ns.jna;

import com.sun.jna.platform.win32.WinBase;
import com.sun.jna.platform.win32.WinNT;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

import java.nio.file.Path;

public class WinFileUtils {
    private static Logger logger = LoggerFactory.getLogger(WinFileUtils.class);

    private final static int GENERIC_READ = 0x80000000;
    private final static int FILE_SHARE_READ = 0x00000001;
    private static final WinBase.SECURITY_ATTRIBUTES SECURITY_ATTRIBUTES = null;
    private static final int OPEN_EXISTING = 3;
    private static final int FILE_FLAG_BACKUP_SEMANTICS = 0x02000000;
    public static final String IO_ERROR = "_ERROR_";

    public static String getUniqueFileId(Path file) {
        Kernel32.BY_HANDLE_FILE_INFORMATION nfo = new Kernel32.BY_HANDLE_FILE_INFORMATION();
        WinNT.HANDLE handle = Kernel32.INSTANCE.CreateFile(file.toString(), GENERIC_READ, FILE_SHARE_READ,
            SECURITY_ATTRIBUTES, OPEN_EXISTING, FILE_FLAG_BACKUP_SEMANTICS, null);
        int errorRet = Kernel32.INSTANCE.GetLastError();
        String identifier;
        if (errorRet != 0) {
            logger.error("創(chuàng)建/打開文件時發(fā)生錯誤 錯誤級別:{} 文件路徑:{}", errorRet, file);
            identifier = IO_ERROR;
        } else {
            Kernel32.INSTANCE.GetFileInformationByHandle(handle, nfo);
            errorRet = Kernel32.INSTANCE.GetLastError();
            if (errorRet != 0) {
                logger.error("獲取文件信息時發(fā)生錯誤 錯誤級別:{} 文件路徑:{}", errorRet, file);
                identifier = IO_ERROR;
            } else {
                identifier = nfo.nFileIndexHigh + nfo.nFileIndexLow.toString() + Integer.toHexString(nfo
                    .dwVolumeSerialNumber.intValue());
            }
        }
        Kernel32.INSTANCE.CloseHandle(handle);
        return identifier;
    }
}

以上代碼綜合參考了這里和這里
再次強(qiáng)調(diào)下挡爵，以上獲取FileId的代碼不適用于Refs系統(tǒng)竖般，如果想支持的更完善，請使用JNA 4.4.0版本 這個版本里的kernel32有一個方法叫做GetFileInformationByHandleEx 它是支持最新的Refs系統(tǒng)的茶鹃。