Eli,
Couple of things I've done (Not necessarily ideal):
* Use a git repo for configs with branches. Switch when needed.
* Point HADOOP_CONF_DIR to the right conf-dir location (or have
triggers that do these for you). Similar to above I guess.
* Alias hadoop multiple times to use different --config conf dir
opts, and use convenient aliased names.
Taking a brief look at hadoopenv sources (Last update seems to be mid
2011), I think it would work with the stable versions of Hadoop even
today. You may give it a whirl too if its approach attracts you.
On Mon, Mar 26, 2012 at 9:30 PM, Eli Finkelshteyn iefin...@gmail.com wrote:
Hi,
I've recently run into a situation where I'll have access to my local single
node cluster, and development cluster, and a production cluster. What's the
best way to work with all three of these, easily switching between which I
want to use? I found this project
http://code.google.com/p/hadoopenv/wiki/README that seems like it's trying
to address my issue, but it hasn't been updated since 2010. Is this still
the preferred way of managing multiple clusters, or is there something
better (preferably something easy to install, as I'll need to convince
others at my company to use whatever solution I find).
Thanks!
Eli
--
Harsh J