Actually Ubuntu comes out of the box with an entry in the hosts file
(/etc/hosts) that maps the computer name to the loopback address. (btw
I'm not sure if this is specific to Ubuntu) The effect is that all name
lookups from the machine for itself resolve to 127.0.0.1. We've seen
Hadoop daemons on Ubuntu bind their sockets to the loopback address,
which will not accept requests from the outside. Symptoms here are that
the cluster works on a single machine but will not distribute.
Just check your netstat tables for the addresses on which the Hadoop
daemons are bound. They should be external addresses. The solution we
use is to remove the entry in the hosts file.
Jakob Homan wrote:
Ubuntu will work fine. The only to-do item is to make sure Sun's Java
is installed and pointed-to, rather than th Open JDK that Ubuntu ships
with by default.
Jakob Homan
Hadoop at Yahoo!
Dmitry Pushkarev wrote:
We have a cluster of over 40 machines all running ubuntu. As long as
you can
run java hadoop should run just fine.
-----Original Message-----
From: CubicDesign [mailto:[email protected]] Sent: Monday, August
17, 2009 10:23 AM
To: [email protected]
Subject: Ubuntu/Hadoop incompatibilities?
Hi.
I have seen two articles (tutorials) on the Internet saying that
Hadoop is incompatible with Ubuntu.
Now I also have problems making Hadoop work and I start to believe
those two articles.
Anybody can confirm that has successfully installed Hadoop on Ubuntu?
Thanks
