hadoop - secondary name node functionality -
can explain words in bold mean taken text book? "state of secondary namenode lags of primary " mean?
secondary name node keeps copy of merged namespace image, can used in event of namenode failing. **however, state of secondary namenode lags of primary, in event of total failure of primary, data loss certain.**the usual course of action in case copy namenode’s metadata files on nfs secondary , run new primary.
thanks in advance
hadoop 1.x:
when start ha hadoop cluster creates file system image keeps metadata information of entire hadopp cluster. when new entry comes hadoop cluster goes edits log. secondary namenode periodically reads , query edits , retrieve information , merge information fsimage. in case namenode fails, hadoop administrator can start hadoop cluster of fsimage , edits.(during start namenode reads edits , fsimage there wont data loss)
fsimage , edits log keeps updated information file system in form of metadata in case of total failure of primary hadoop administrator can recover cluster information of edits log , fsimage.
hadoop 2.x:
in hadoop 1.x namenode single point of failure. failure of namenode downtime entire hadoop cluster. planned maintenance events such software or hardware upgrades on namenode machine result in periods of cluster downtime.to overcome issue hadoop community added high availability feature. during setting of hadoop cluster can choose type of cluster want.
the hdfs namenode high availability feature enables run redundant namenodes in same cluster in active/passive configuration hot standby.both namenode require same type of hardware configuration.
in ha configuration 1 namenode active , other in standby state.the zkfailovercontroller (zkfc) zookeeper client monitors , manages state of namenode. when active namenode goes down, makes standby active namenode, , primary namenode become standby when start them. please can more on on website: http://docs.hortonworks.com/hdpdocuments/hdp2/hdp-2.0.8.0/bk_system-admin-guide/content/ch_hadoop-ha-5.html
in ha hadoop cluster active namenode reads , write metadata information in journalnode(quorum-based storage only). journalnode separate node in ha hadoop cluster used reads , write edits log , fsimage.
standby namenodealways synchronized active namenode, both communicate each other through journal node. when namespace modification performed active node, durably logs record of modification majority of these jns. standby namenode monitors edit logs @ journal nodes , updates namespace accordingly.in event of failover, standby namenode ensure namespace updated according edit logs before changes active state. when standby in active state start writing edits log journalnode.
hadoop don't keep data namenode, data resides in datanode, in case of namenode failure there wont loss of data.
Comments
Post a Comment