Hbase bulkload源代码阅读笔记

1.LoadIncrementalHFiles.doBulkLoad(Path hfofDir, HTable table)

首先用discoverLoadQueue方法扫描出hfofDir下有哪些fhile文件，再循环调用tryLoad方法把每个文件load进去，这是一个串行的过程。

Deque<LoadQueueItem> queue = null;
queue = discoverLoadQueue(hfofDir);
while (!queue.isEmpty()) {
LoadQueueItem item = queue.remove();
tryLoad(item, conn, table.getTableName(), queue);
}

2.LoadIncrementalHFiles.discoverLoadQueue(Path hfofDir)

hfofDir下是两层目录结构family-->hfile，因此二重循环遍历每个hfile文件，加到Deque里返回，hfofDir下以"_"开头的不是family目录。
LoadQueueItem的数据结构用于记录family和hfile
final byte[] family;
final Path hfilePath;

FileStatus[] familyDirStatuses = fs.listStatus(hfofDir);//第一层目录
Deque<LoadQueueItem> ret = new LinkedList<LoadQueueItem>();
for (FileStatus stat : familyDirStatuses) {
Path familyDir = stat.getPath();
// Skip _logs, etc
if (familyDir.getName().startsWith("_")) continue;//以"_"开头的不是family目录。
byte[] family = familyDir.getName().getBytes();
Path[] hfiles = FileUtil.stat2Paths(fs.listStatus(familyDir));//第二层目录
for (Path hfile : hfiles) {
if (hfile.getName().startsWith("_")) continue;
ret.add(new LoadQueueItem(family, hfile));
}
}

3.LoadIncrementalHFiles.tryLoad(final LoadQueueItem item,
HConnection conn, final byte[] table,
final Deque<LoadQueueItem> queue)

首先检查当前的hfile所属的region是否已经发生分裂，如果发生分裂，则将hfile分裂成匹配新region的两个hfile，并将这两个hfile放入deque；哪果没有发生分裂，则调用region所在server的bulkLoadHFile方法将hfile导入。重点是以下几句

if (!hri.containsRange(first, last)) {//判断包含firstkey的当前region是否包含hfile的startkey和endkey，如果不包含说明当前region是分裂过的
LOG.info("HFile at " + hfilePath + " no longer fits inside a single " +
"region. Splitting...");

HColumnDescriptor familyDesc = hri.getTableDesc().getFamily(item.family);
Path botOut = new Path(tmpDir, hri.getEncodedName() + ".bottom");
Path topOut = new Path(tmpDir, hri.getEncodedName() + ".top");
splitStoreFile(getConf(), hfilePath, familyDesc, hri.getEndKey(),
botOut, topOut);//以当前region的endkey为中值分裂hfile为两个，文件存为.bottom和.top

// Add these back at the *front* of the queue, so there's a lower
// chance that the region will just split again before we get there.
//.bottom和.top重新放回queue
queue.addFirst(new LoadQueueItem(item.family, botOut));
queue.addFirst(new LoadQueueItem(item.family, topOut));
LOG.info("Successfully split into new HFiles " + botOut + " and " + topOut);
return null;
}

byte[] regionName = location.getRegionInfo().getRegionName();
server.bulkLoadHFile(hfilePath.toString(), regionName, item.family);//如果包含，直接调用region所在server的bulkLoadHFile方法将hfile导入
return null;

4.LoadIncrementalHFiles.splitStoreFile(
Configuration conf, Path inFile,
HColumnDescriptor familyDesc, byte[] splitKey,
Path bottomOut, Path topOut)

//以splitKey为中值，将inFile拷贝分裂为bottomOut和topOut两个文件
Reference topReference = new Reference(splitKey, Range.top);
Reference bottomReference = new Reference(splitKey, Range.bottom);

copyHFileHalf(conf, inFile, topOut, topReference, familyDesc);
copyHFileHalf(conf, inFile, bottomOut, bottomReference, familyDesc);

5.HRegionServer.bulkLoadHFile(String hfilePath, byte[] regionName,
byte[] familyName)

这是个HRegionInterface下的远程调用，是在regionserver中执行的。

checkOpen();//检查region是否已停，已经停了便不再导
HRegion region = getRegion(regionName);//从regionserver中拿到region
region.bulkLoadHFile(hfilePath, familyName);//这步才开始导

6.HRegion.bulkLoadHFile(String hfilePath, byte[] familyName)
throws IOException {
startRegionOperation();//上读锁
try {
Store store = getStore(familyName);
if (store == null) {
throw new DoNotRetryIOException(
"No such column family " + Bytes.toStringBinary(familyName));
}
store.bulkLoadHFile(hfilePath);//调store的同名方法
} finally {
closeRegionOperation();//解读锁
}

}

7.Store.bulkLoadHFile(String srcPathStr)

就三步：首先将hfile重命名到store的目录下；其次将hfile包装成StoreFile对象装载到Store的列表里。在这两步之前是再一次检查region的startkey和endkey是否跟hfile的匹配

//再次检查是否匹配region
HRegionInfo hri = region.getRegionInfo();
if (!hri.containsRange(firstKey, lastKey)) {
throw new WrongRegionException(
"Bulk load file " + srcPathStr + " does not fit inside region "
+ this.region);
}

//挪文件
Path srcPath = new Path(srcPathStr);
Path dstPath = StoreFile.getRandomFilename(fs, homedir);
LOG.info("Renaming bulk load file " + srcPath + " to " + dstPath);
StoreFile.rename(fs, srcPath, dstPath);


//装载文件以提供在线服务
// Append the new storefile into the list
this.lock.writeLock().lock();//加store的写锁
try {
ArrayList<StoreFile> newFiles = new ArrayList<StoreFile>(storefiles);
newFiles.add(sf);
this.storefiles = sortAndClone(newFiles);
notifyChangedReadersObservers();
} finally {
this.lock.writeLock().unlock();//解store的写锁
}