zookeeper的maxSessionTimeout默认值导致hbase regionserver超时

zookeeper的maxSessionTimeout默认值导致hbase regionserver超时





在hbase中经常会遇到regionserver挂掉的情况，查看日志会看到这样的错误信息

2016-02-16 11:51:24,882 WARN  [master/hadoop02/192.168.100.5:6000] zookeeper.ZKUtil: master:6000-0x3525df55101a556, quorum=hadoop02:2181,hadoop03:2181,hadoop01:2181, baseZNode=/hbase Unable to get data of znode /hbase/xx

org.apache.zookeeper.KeeperException$SessionExpiredException: KeeperErrorCode = Session expired for /hbase/xx

        at org.apache.zookeeper.KeeperException.create(KeeperException.java:127)

        at org.apache.zookeeper.KeeperException.create(KeeperException.java:51)

        at org.apache.zookeeper.ZooKeeper.getData(ZooKeeper.java:1155)

        at org.apache.hadoop.hbase.zookeeper.RecoverableZooKeeper.getData(RecoverableZooKeeper.java:359)

        at org.apache.hadoop.hbase.zookeeper.ZKUtil.getData(ZKUtil.java:745)







说明与zookeeper的连接的session超时了，可是在hbase-site.xml中设置了zookeeper的超时时间为2分钟，原来

zookeeper中自带两个参数设置了session的超时时间，在启动时日志中会显示

2016-02-16 11:39:49,164 - INFO  [main:QuorumPeer@992] - tickTime set to 2000

2016-02-16 11:39:49,165 - INFO  [main:QuorumPeer@1012] - minSessionTimeout set to -1

2016-02-16 11:39:49,165 - INFO  [main:QuorumPeer@1023] - maxSessionTimeout set to -1





而minSessionTimeout 和maxSessionTimeout 是用下面的方式算出来的

public int getMinSessionTimeout()

{ return minSessionTimeout == -1 ? tickTime * 2 : minSessionTimeout; } 

public int getMaxSessionTimeout() 

{ return maxSessionTimeout == -1 ? tickTime * 20 : maxSessionTimeout; }





默认情况，tickTime=2sec，那么minSessionTimeout 和 maxSessionTimeout 分别是4sec和40sec





所以在hbase中设置超时时间是没用的，必须修改zookeeper自身的maxSessionTimeout为1200000，才能真正起到加长zookeeper的session超时时间的作用