Hi all, I am trying to setup my cluster to move the secondary namenode into a different machine. I would like to know how to set it up in details. Cluster setup: hm01 (master) hm02 (backup master) hd01 - hdxx (slaves)
Here is what I know. 1) create a file under conf in hm01, call it secondaryNameNode, and place hm02's ip address in it. 2) add the hm02's ip and port to dfs.secondary.http.address in the xml file 2) run hadoop-daemons.sh --config $HADOOP_CONF_DIR --hosts SecondaryNameNode start secondarynamenode The questions is, how does hm01 and hm02 communicate with each other and the mechanism works? do I ran step 2 and 3 on hm01 or hm02? How does the secondaryNameNode get the edits from hm01 and sents back the merged fsimages to hm01. Thanks in advance if someone can explain these things to me in details. Felix
