Hi, Agree that would be great to do since its an incompatible change (going forward). Would you like to contribute a doc fix for this?
I do believe the other (client-oriented) components in the Hadoop ecosystem, such as Pig, Hive, etc. can reuse HADOOP_HOME/etc. if set, for some benefits (such as reading discovering the right cluster, etc.). I'd say its primarily user benefit. P.s. For 1.x, you may also want to instead use packages offered via Apache Bigtop: https://cwiki.apache.org/confluence/display/BIGTOP/How+to+install+Hadoop+distribution+from+Bigtop, which offers an integration-tested stack of ecosystem components (i.e. proper, working version bundles). On Thu, Jun 14, 2012 at 2:34 PM, Ondřej Klimpera <klimp...@fit.cvut.cz> wrote: > Thanks, for your reply. It would be great to mention this in your tutorial > on your web sites. Is the name of the HADOOP_PREFIX/HOME/INSTALL crucial to > Hadoop, or it's just user benefit to set this variable. > > Thanks for reply. > > > On 06/14/2012 07:46 AM, Harsh J wrote: >> >> Hi Ondřej, >> >> Due to a new packaging format, the Apache Hadoop 1.x has deprecated >> the HADOOP_HOME env-var in favor of a new env-var called >> 'HADOOP_PREFIX'. You can set HADOOP_PREFIX, or set >> HADOOP_HOME_WARN_SUPPRESS in your environment to a non-empty value to >> suppress the warning. >> >> On Thu, Jun 14, 2012 at 11:11 AM, Ondřej Klimpera<klimp...@fit.cvut.cz> >> wrote: >>> >>> Hello, why when running Hadoop, there is always HADOOP_HOME shell >>> variable >>> being told to be deprecated. How to set installation directory on cluster >>> nodes, which variable is correct. >>> >>> Thanks >>> >>> Ondrej Klimpera > > < -- Harsh J
