To make sure we have the same context, I think you're working through the bulk ingest overview's example. Please correct me if that's wrong. Before running any of the accumulo examples, you need to do some user set up. None of them should be run as root nor any of the service principles (accumulo, hdfs, etc). The user that will run the data generation needs to be able to run MapReduce jobs. See the full docs for instructions on provisioning such a user. In short, ensure they have a user account on all worker nodes and that they have a user directory in HDFS (creating said home directory will require action as the hdfs super user). The user you created above will be used for the data generation step. If you are running on a secure cluster, you will need to use your kerberos password before submitting the job. Otherwise the generation step only requires an initial local login. The data loading step requires an Accumulo user. You should create a user via the Accumulo shell. Be sure to replace the instance name, zookeeper servers, and user/password given in the ARGS line with ones appropriate for your cluster. This loading should not be done as the Accumulo root user. Let me know if you have any further problems.
... View more
