Skip to content

Latest commit

 

History

History
20 lines (12 loc) · 678 Bytes

README.md

File metadata and controls

20 lines (12 loc) · 678 Bytes

hadoop-autobuild

This is a hacky way of getting a hadoop cluster quickly up on EC2.

First up, spin up nodes and pick one as a master.

Checkout this code on the master and create a file named "slaves" filled with the other hostnames (or for a single node cluster, you can say "localhost" in it).

run "make install" to download & compile hadoop/branch-2

then "make start" to start the daemons on all the nodes correctly.

You can edit gen-conf.py to add more config options and play around with the configs.

There are 3 assumptions made on this

  1. all nodes are identical
  2. you have root everywhere
  3. root is passphraseless (or at least, agent handled)