Evert Lammerts 2012-08-08, 10:10
DistributedCache.getLocalCacheFiles does not seem to work correctly in 0.20.205.0. In my Mapper I need a local filesystem path on the cluster nodes. getLocalCacheFiles is not populated though, and always returns null. Calling getCacheFiles in my Mapper gives me the URI's of the cacheFiles in the conf object, which shows DistributedCache.addCacheFile at least added the files to the configuration. The file I'm trying to get access to through the DistributedCache is alvailable in HDFS - a dfs -ls confirms this.
In see some other posts related to this issue (a recent one in https://issues.apache.org/jira/browse/MRUNIT-124), but no workarounds or solutions.
- Evert Lammerts