Some node sizes are not started with enough disk space to run TPC-H testing

m2.small works because the AMI for that size has a root device at `/dev/xvda`, which the CloudFormation populates. However, the AMI for m3.medium uses a root device of `/dev/sda`, which the CloudFormation does not populate, so the OS does not expose enough space for us to do TPC-H testing.

Start by locating the CloudFormation configuration that provisions the m2.small and m3.medium nodes, then compare how each AMI root device is populated. Reproduce the node setup and run the TPC-H testing workflow. Done means both node sizes expose enough disk space for TPC-H testing.

Written by the indexing model from the issue text.