On Wed, Nov 20, 2013 at 11:41 PM, Jukka Zitting <jukka.zitt...@gmail.com> wrote: > Yes, sounds like a good candidate. The only additional bit we'd need > is a timestamp that allows indexers on other cluster nodes to > automatically resume processing if an active indexing task dies for > whatever reason without a chance to clear the flag.
Implemented such a logic with OAK-1246 Chetan Mehrotra
