Terrier 3.6 ships with the Jar files for Hadoop 0.20 (CDH3) (i.e. hadoop-0.20.2+228). However, it is possible to upgrade Terrier to work with more recent Hadoop clusters.

1. Delete lib/hadoop-0.20*.jar from Terrier's lib/ directory.

2. Remove lib/hadoop/ directory from Terrier's lib/ directory.

3. Copy the hadoop-*.jar files from your Hadoop distribution to Terrier's lib/ directory.

4. Make a hadoop/ directory in Terrier's lib/ directory.

5. Copy all jar files from Hadoop's lib/ directory to the lib/hadoop/ directory.

We welcome feedback as to which versions of Hadoop this works with.

last edited 2014-04-04 12:13:40 by 0402393m