# Tribe on\_conflict **URL:** \ **Category:** Elasticsearch\ **Created:** [October 21, 2015, 10:38pm UTC](https://discuss.elastic.co/t/tribe-on-conflict/32727 "2015-10-21T22:38:49Z")\ **Posts on this page:** 2\ **Page:** 1
**Author:** ![Bobby\_Pilot](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/bobby_pilot/32/1186_2.png) [@Bobby\_Pilot](https://discuss.elastic.co/u/Bobby_Pilot)\ **Post date:** [October 21, 2015, 10:38pm UTC](https://discuss.elastic.co/t/tribe-on-conflict/32727/1 "2015-10-21T22:38:49Z")
I see that tribe.on\_conflict can be set to prefer a specific tribe. I'm curious if that is tied to zen discovery in any way? In an HA mult-datacenter failover scenario, I'm wondering if we can rely on a federated tribe cluster to favor its local cluster indices when all is well but failover to the other tribe when the conflict no longer exists. How often is the conflict checked? Here is the scenario I'm rationalizing... Cluster A is in datacenter A and ClusterB is datacenter B. They are mirror images using some kind of data mirroring/ replication strategy (another problem to solve 😉 ) . In each data center is a few tribe nodes the join both clusterA and clusterB across the data centers. Since the clusters are mirrored, obviously every index will conflict. The tribes would be configured to favor their data centers cluster. So tribeA would on\_conflict prefer its indexes and tribeB its indexes. If tribeA loses its local cluster (maintenance, failure, etc), will the tribe know to defer to the remote tribeB's indices that no longer conflict? Make sense? ---
**Author:** ![system](https://us1.discourse-cdn.com/elastic/original/3X/1/a/1ac57faf039f6b580b3f104ef42a2a89e41014de.png) [@system](https://discuss.elastic.co/u/system)\ **Post date:** [July 5, 2017, 11:43pm UTC](https://discuss.elastic.co/t/tribe-on-conflict/32727/2 "2017-07-05T23:43:24Z")