Bryan A. Pendleton wrote:
I had to fix things up because there's not equivalent of hadoop-site.xml for environments and slaves - why use a directory-dependent set of configuration files, but ~/.slaves by default?
No good reason. The default slaves file should probably be something like conf/hadoop-slaves.txt. Please file a bug and/or patch.
Anyway, once I got JAVA_HOME fixed up and started a single set of daemons, I still get the 404 error.
Are there any errors in the jobtracker log file, when it starts Jetty?
Actually, I think I'll go file a feature request to this effect, but it'd be nice if hadoop/nutch specific environment variables were stored in someplace in conf, as well.
I agree that this should be done. My plan is to add a hadoop-env.sh file that is sourced by hadoop scripts. This is where folks can set JVM and HADOOP-specific environment stuff. Note that, in a distributed configuration, this file, like hadoop-site.xml and the hadoop code, has to be copied to all nodes as a bootstrapping step.
Doug
