I am not sure if Benjamin is using Cloudera Manager or if he is using the Apache Distribution. However, I thought it might help to see the two files and configurations that should be focused on in regard to making Hadoop jars available to accumulo.
From what it sounds like, it would seem that either the firewall is blocking communication with the tablet servers, or the hosts that have Accumulo tablets (slaves) do not have Hadoop client configuration setup properly. I would make sure the user account for the tserver process can see the exact same filesystem as the master where the shell init was run. I have seen ‘hadoop fs -ls’ actually show the local linux filesystem when the /etc/hadoop/hdfs-site.xml and mapred-site.xml are not configured properly. From: [email protected] [mailto:[email protected]] On Behalf Of Sean Busbey Sent: Tuesday, March 18, 2014 3:04 PM To: Accumulo User List Subject: Re: Installing with Hadoop 2.2.0 On Tue, Mar 18, 2014 at 1:37 PM, Ott, Charlie H. <[email protected]<mailto:[email protected]>> wrote: I posted my config here: http://championofcyrodiil.blogspot.com/2014/03/installing-accumulo-150-on-cloudera.html Hi Charlie! Do you mean Benjamin is using this same configuration? From his previous postings I thought he was using the Apache Hadoop 2.2.0 release and not a vendor specific repackaging. -Sean
