You need additional settings to make ResourceManager auto-failover.
http://hadoop.apache.org/docs/r2.4.1/hadoop-yarn/hadoop-yarn-site/ResourceManagerHA.html JobHistoryServer does not have automatic failover feature. Regards, Akira (2014/08/05 20:15), arthur.hk.c...@gmail.com wrote:
Hi I have set up the Hadoop 2.4.1 with HDFS High Availability using the Quorum Journal Manager. I am verifying Automatic Failover: I manually used “kill -9” command to disable all running Hadoop services in active node (NN-1), I can find that the Standby node (NN-2) now becomes ACTIVE now which is good, however, the “ResourceManager” service cannot be found in NN-2, please advise how to make ResourceManager and JobHistoryServer auto-failover? or do I miss some important setup? missing some settings in hdfs-site.xml or core-site.xml? Please help! Regards Arthur BEFORE TESTING: NN-1: jps 9564 NameNode 10176 JobHistoryServer 21215 Jps 17636 QuorumPeerMain 20838 NodeManager 9678 DataNode 9933 JournalNode 10085 DFSZKFailoverController 20724 ResourceManager NN-2 (Standby Name node) jps 14064 Jps 32046 NameNode 13765 NodeManager 32126 DataNode 32271 DFSZKFailoverController AFTER NN-1 dips 17636 QuorumPeerMain 21508 Jps NN-2 jps 32046 NameNode 13765 NodeManager 32126 DataNode 32271 DFSZKFailoverController 14165 Jps
