my mistake:
Machine 1: NameNode (Master) and SecondaryNameNode (Master)
Machine 2: ResourceManager (Master) and ProxyServer (Master)
Machine 3,4,5: DataNode (Slave) and NodeManager (Slave)
On 09.08.2014 11:37, Alexander Sirotin wrote:
Hi,
I am not an expert, but I am running a 5 machine cluster.
I do the following.
Machine 1: DataNode and NodeManager
Machine 2: ResourceManager (Yarn-Master) and ProxyServer
Machine 3,4,5: NameNode and SecondaryNameNode
If you like you can also start the slave-processes also on the
master-machines, so you are also using the diskspace of these. For me
it is not neccessary yet and I am using the master-nodes dedicatedly.
To define you just put into the hostname or ip-address into
etc/hadoop/[masters|slaves]. 1 host per line
Best regards
On 09.08.2014 02:30, xeniad20 wrote:
Hi experts,
What is the best for multinode Hadoop configuration in a small
cluster with 4 machines only.
How can we define master, secondarynamenode and datanodes in the
/hadoop/conf/masters and slaves files?
Thanks