> On Mar 14, 2016, at 10:31, Mich Talebzadeh <mich.talebza...@gmail.com> wrote: > > That is an interesting point Alan. > > Does this imply that Hive on Spark (Hive 2 encourages Spark or TEZ) is going > to have an issue with transactional tables? I was partially wrong. In HIVE-12366 Wei refactored this so that the heartbeats are sent from the client rather than from the execution engine. This JIRA has been committed to master and branch-1, but after 1.2. So Hive on Spark should work fine with transactional tables in Hive 2.x. In 1.2 and earlier it will not.
Alan.
