Hi Brian,
I just tested this functionality in a small cluster at my disposal. It
seems that you need not specify the namenode hostname in the
conf/masters file. Just specifying the secondary namenode hostname in
the conf/masters file will do the job for you. To put it formally,
- if your cluster is already running, then shut it down.
- modify the masters file in conf directory to have only the hostname of
the machine where you would start up your secondary namenode.
- run a ./bin/start-all.sh
once the startup completes, you can check with jps whether namenode and
secondary namenodes started up in respective hosts.
hope this helps.
Pratyush
Pratyush Banerjee wrote:
Hi All,
I am facing the same confusions regarding setting up secondary
namenode server in a separate machine. It would be very helpful if
anyone can help us in the direction.
Pratyush
Brian Karlak wrote:
Hello All --
We have a 20-node cluster running 0.17.1. Currently, the secondary
namenode process is running on the same machine as the primary
namenode. We would like to move it to a separate machine in the
cluster, as recommended. However, I cannot seem to find much in the
way of documentation on how to do this properly.
I found:
http://hadoop.apache.org/core/docs/current/hdfs_user_guide.html#Secondary+Namenode
It is usually run on a different machine than the primary Namenode
since its memory requirements are on the same order as the primary
namemode. The secondary namenode is started by bin/start-dfs.sh on
the nodes specified in conf/masters file.
as well as the "dfs.secondary.http.address" property in
hadoop-default.xml file.
However, this seems a bit confusing. What is it in conf/masters that
specifies that one hostname should be a master, and the other a
secondary?
Many thanks in advance for any guidance anyone can give.
Brian
