工作中运行hbase报的zookeeper异常信息：
2013-06-28 18:26:59,946 WARN org.apache.hadoop.hbase.zookeeper.ZKUtil: master:60000-0x33f7eb9d0650002-0x33f7eb9d0650002-0x33f7eb9d0650002 Unable to get data of znode /hbase/master
org.apache.zookeeper.KeeperException$SessionExpiredException: KeeperErrorCode = Session expired for /hbase/master
    at org.apache.zookeeper.KeeperException.create(KeeperException.java:118)
    at org.apache.zookeeper.KeeperException.create(KeeperException.java:42)
    at org.apache.zookeeper.ZooKeeper.getData(ZooKeeper.java:927)
    at org.apache.hadoop.hbase.zookeeper.ZKUtil.getDataInternal(ZKUtil.java:577)
    at org.apache.hadoop.hbase.zookeeper.ZKUtil.getDataAndWatch(ZKUtil.java:554)
    at org.apache.hadoop.hbase.zookeeper.ZKUtil.getDataAsAddress(ZKUtil.java:648)
    at org.apache.hadoop.hbase.master.ActiveMasterManager.stop(ActiveMasterManager.java:202)
    at org.apache.hadoop.hbase.master.HMaster.run(HMaster.java:318)
2013-06-28 18:26:59,946 ERROR org.apache.hadoop.hbase.zookeeper.ZooKeeperWatcher: master:60000-0x33f7eb9d0650002-0x33f7eb9d0650002-0x33f7eb9d0650002 Received unexpected KeeperException, re-throwing exception
org.apache.zookeeper.KeeperException$SessionExpiredException: KeeperErrorCode = Session expired for /hbase/master
    at org.apache.zookeeper.KeeperException.create(KeeperException.java:118)
    at org.apache.zookeeper.KeeperException.create(KeeperException.java:42)
    at org.apache.zookeeper.ZooKeeper.getData(ZooKeeper.java:927)
    at org.apache.hadoop.hbase.zookeeper.ZKUtil.getDataInternal(ZKUtil.java:577)
    at org.apache.hadoop.hbase.zookeeper.ZKUtil.getDataAndWatch(ZKUtil.java:554)
    at org.apache.hadoop.hbase.zookeeper.ZKUtil.getDataAsAddress(ZKUtil.java:648)
    at org.apache.hadoop.hbase.master.ActiveMasterManager.stop(ActiveMasterManager.java:202)
    at org.apache.hadoop.hbase.master.HMaster.run(HMaster.java:318)
2013-06-28 18:26:59,947 ERROR org.apache.hadoop.hbase.master.ActiveMasterManager: master:60000-0x33f7eb9d0650002-0x33f7eb9d0650002-0x33f7eb9d0650002 Error deleting our own master address node
org.apache.zookeeper.KeeperException$SessionExpiredException: KeeperErrorCode = Session expired for /hbase/master
    at org.apache.zookeeper.KeeperException.create(KeeperException.java:118)
    at org.apache.zookeeper.KeeperException.create(KeeperException.java:42)
    at org.apache.zookeeper.ZooKeeper.getData(ZooKeeper.java:927)
    at org.apache.hadoop.hbase.zookeeper.ZKUtil.getDataInternal(ZKUtil.java:577)
    at org.apache.hadoop.hbase.zookeeper.ZKUtil.getDataAndWatch(ZKUtil.java:554)
    at org.apache.hadoop.hbase.zookeeper.ZKUtil.getDataAsAddress(ZKUtil.java:648)
    at org.apache.hadoop.hbase.master.ActiveMasterManager.stop(ActiveMasterManager.java:202)
    at org.apache.hadoop.hbase.master.HMaster.run(HMaster.java:318)

以下异常原因以及解决方法为摘抄，待验证。

异常原因：

hbase中和GC相关的参数：

修改前（默认）：

export HBASE_OPTS="$HBASE_OPTS -ea -verbose:gc -Xloggc:$HBASE_LOG_DIR/hbase.gc.log -XX:ErrorFile=$HBASE_LOG_DIR/hs_err_pid.log -XX:+PrintGCTimeStamps -XX:+PrintGCDetails -XX:+HeapDumpOnOutOfMemoryError -XX:+UseConcMarkSweepGC -XX:+CMSIncrementalMode"

咨询开发修改后：

export HBASE_OPTS="$HBASE_OPTS -verbose:gc -Xloggc:$HBASE_LOG_DIR/hbase.gc.log -XX:ErrorFile=$HBASE_LOG_DIR/hs_err_pid.log -XX:+PrintGCDateStamps -XX:+PrintGCDetails -XX:+HeapDumpOnOutOfMemoryError -XX:+UseConcMarkSweepGC -XX:+UseParNewGC -XX:CMSInitiatingOccupancyFraction=70"

-XX UseConcMarkSweepGC ：设置年老代为并发收集。（新老都有）

老：-XX:+CMSIncrementalMode :设置为增量模式。适用于单CPU情况。

新：-XX:+UseParNewGC：设置年轻代为并行收集。可与 CMS 收集同时使用。

-XX:CMSInitiatingOccupancyFraction=70:这个参数是我觉得产生最大作用的。因为最终的目的是减少FULL GC，因为full gc是会block其他线程的。

默认触发GC的时机是当年老代内存达到90%的时候，这个百分比由 -XX:CMSInitiatingOccupancyFraction=N 这个参数来设置。concurrent mode failed发生在这样一个场景：
当年老代内存达到90%的时候，CMS开始进行并发垃圾收集，于此同时，新生代还在迅速不断地晋升对象到年老代。当年老代CMS还未完成并发标记时，年老 代满了，悲剧就发生了。CMS因为没内存可用不得不暂停mark，并触发一次全jvm的stop the world（挂起所有线程），然后采用单线程拷贝方式清理所有垃圾对象，也就是full gc。而我们的bulk的最开始的操作就是各种删表,建表频繁的操作，就会使用掉大量master的年轻代的内存，就会发生上面发生的场景，发生full gc。

解决办法：CMSInitiatingOccupancyFraction=70表示年老代占到约70%时就开始执行CMS，这样就不会出现（或很少出现）Full GC了。