ClusterDumper writes to System.out or local filesystem only (I would like to
write to s3 when running on Elastic MapReduce)
---------------------------------------------------------------------------------------------------------------------------
Key: MAHOUT-701
URL: https://issues.apache.org/jira/browse/MAHOUT-701
Project: Mahout
Issue Type: Bug
Components: Utils
Environment: Amazon Elastic MapReduce
Reporter: Dave Lewis
Priority: Minor
At the end of a kmeans job at EMR, I like to look through the clusters.
Unfortunately ClusterDumper writes to System.out or a local file. I added a
small conditional to examine the filename to see if it starts with s3n://, and
if so to open up a FileSystem to write to s3 instead of to the local filesystem
so that the output file is available after the cluster is shut down. I am
creating the patch now (tests are still running from the change in MAHOUT-700),
so I will add the patch to this issue shortly.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
