HBase Split 过程
看了下hbase split的代码 记录一下学习心得
private void split(final HRegion parent, final byte [] midKey)throws IOException { final long startTime = System.currentTimeMillis(); SplitTransaction st = new SplitTransaction(parent, midKey); // If prepare does not return true, for some reason -- logged inside in // the prepare call -- we are not ready to split just now. Just return. if (!st.prepare()) return; try { st.execute(this.server, this.server);
SplitTransation表示处理split是一个类似事务性质的行为,接下来我们会经常和它打交道
初始化了一个path:tablename/parent/split/
st.prepare():构造了两个HRegionInfo:hri_a和hri_b,a是top,其startkey为parent的startkey,endkey是midkey;hri_b是bottom,startkey为midkey,endkey是parent的endkey。
然后st.execute(this.server, this.server);
1.创建splitdir:tablename/parent/split/,在状态机中加入CREATE_SPLIT_DIR
createSplitDir(this.parent.getFilesystem(), this.splitdir); this.journal.add(JournalEntry.CREATE_SPLIT_DIR);
2.close parentregion,并返回所有的storefile,状态机加入CLOSED_PARENT_REGION
close的过程会刷一次磁盘memstore的数据写入磁盘;
List<StoreFile> hstoreFilesToSplit; hstoreFilesToSplit = this.parent.close(false);
3.从rs的onlineregion列表中移除parent region,状态机添OFFLINED_PARENT
services.removeFromOnlineRegions(this.parent.getRegionInfo().getEncodedName()); this.journal.add(JournalEntry.OFFLINED_PARENT);
4.splitStoreFiles(this.splitdir, hstoreFilesToSplit);
多线程处理所有的storefiles.创建文件/table/parent/split/hri_a
(hri_b)/family/storefilename. parent,在该文件中写入boolean top和midkey
splitStoreFiles(this.splitdir, hstoreFilesToSplit); this.journal.add(JournalEntry.STARTED_REGION_A_CREATION);
5.状态机添STARTED_REGION_A_CREATION,创建子Region A。将上面的hri_a下的文件move到table/hri_a下即变为/hbase/hri_a/family/storefilename.parent
6.同理状态机添STARTED_REGION_B_CREATION而后创建子Region B
this.journal.add(JournalEntry.STARTED_REGION_A_CREATION); HRegion a = createDaughterRegion(this.hri_a,this.parent.flushRequester, this.parent.rsServices); this.journal.add(JournalEntry.STARTED_REGION_B_CREATION); HRegion b = createDaughterRegion(this.hri_b, this.parent.flushRequester, this.parent.rsServices);
7.meta表中下线parent;将parent在meta表中offline和split标志置为true,parent
Region添加两列SPLITA和SPLITB,值为HRegioninfo
8.状态机添PONR
try{ if (!testing) { MetaEditor.offlineParentInMeta(server.getCatalogTracker(), this.parent.getRegionInfo(), a.getRegionInfo(), b.getRegionInfo()); } } catch(IOException e){ throw e; } finally{ this.journal.add(JournalEntry.PONR); }
9.开两个线程open REGION A和B。
1>调用openRegion函数进行initilize,主要步骤如下
a)向hdfs上写入.regionInfo文件以便meta挂掉以便恢复
b)初始化其下的HStore,主要是LoadStoreFiles函数:
对于该store函数会构造storefile对象,从hdfs上获取路径和文件,每个文件一个
storefile对象,对每个storefile对象会读取文件上的内容创建一个
HalfStoreFileReader读对象来操作该region的父region上的相应的文件,及该
region上目前存储的是引用文件,其指向的是其父region上的相应的文件,对该
region的所有读或写都将关联到父region上
2>将子Region添加到rs的online region列表上,并添加到meta表上
DaughterOpener aOpener = new DaughterOpener(server, services, a); DaughterOpener bOpener = new DaughterOpener(server, services, b); aOpener.start(); bOpener.start(); try { aOpener.join(); bOpener.join(); } catch (InterruptedException e) { Thread.currentThread().interrupt(); throw new IOException("Interrupted " + e.getMessage()); }
整个过程如上所述,如果碰到异常的话,会进行rollback检查状态机上的状态并依次进行rollback