Windows only or extension to Linux

I reviewed the MSI parameters and noticed that the NAMENODE_HOST is a mandatory field to configure. How does it work if no Hadoop instances (name, jobtracker) are configured? Do I first need to configure a (name and jobtracker) Hadoop instance and then configure the MSI for the remaining task- and datanodes? Or can the MSI also be used to configure the first instances? The manual doesn’t say anything about this.

Topic

[Larry LiuModerator]

Hi, Piethein

You are right about the clusterproperties.txt. Once the installation is completed, each node knows the configuration. The purpose of clusterproperties.txt is to tell the installer the layout of the cluster and configure each node accordingly.

Thanks

Larry

[Piethein StrengholtMember]

@tedr

Thanks for your answer. So how does it work when I want to install a namenode host and mutiple data nodes? One time configure the clusterproperties.txt, run the MSI on all machines?

Is it correct that the MSI will read the clusterproperties.txt, detect and then decide to configure the machine as a namenode or datanode for instance, based on the servername and information retrieved from the clusterproperties.txt?

[tedrMember]

Hi Piethein,

Thanks for looking at HDP for Windows.

The NAMENODE_HOST variable is not meant to point at an existing instance of an hadoop namenode, but insteat it should point at the host you wish to become the namenode during cluster install.

Thanks again,
Ted.

[Author]

Replies