This recent article from David Chaiken at Altiscale discusses how to debug Hive (Hadoop) through an anecdote regarding a customer's struggling Hive job. According to Chaiken, there are downsides to working with Hadoop - "[it] is powerful," he says, "and it’s experiencing a tremendous rate of innovation, but it also has many rough edges" - and sometimes Hadoop does not offer a lot of information in terms of what has gone wrong.

The error in Chaiken's anecdote is eventually solved, and it's a simple mistake that snowballs into a bigger problem because of some of these rough edges in Hadoop, but Chaiken's walk-through of his solution provides an interesting look at debugging a Hive job when the problem is unclear. Take a look at the full article and find out one way to approach a Hive job that has lost its way.