Hi,
I've run into the following issue running nexus in docker (3.1) which has its '/nexus-data' on AWS EFS share.
2016-11-14 17:19:31,960+0000 WARN [elasticsearch[84D7013A-6E76DF96-CD56BA97-EDE2864B-01E6A584][management][T#2]] *SYSTEM org.elasticsearch.cluster.routing.allocation.decider - [84D7013A-6E76DF96-CD56BA97-EDE2864B-01E6A584] high disk watermark [90%] exceeded on [Xe5R7PlYSd-RVoZ3GiBsPg][84D7013A-6E76DF96-CD56BA97-EDE2864B-01E6A584][/nexus-data/elasticsearch/nexus/nodes/0] free: 8191.9pb[-99.9%], shards will be relocated away from this node
2016-11-14 17:19:31,962+0000 INFO [elasticsearch[84D7013A-6E76DF96-CD56BA97-EDE2864B-01E6A584][management][T#2]] *SYSTEM org.elasticsearch.cluster.routing.allocation.decider - [84D7013A-6E76DF96-CD56BA97-EDE2864B-01E6A584] rerouting shards: [high disk watermark exceeded on one or more nodes]
EFS shows up as 8 Exabytes to a client system, it seems like elasticsearch 2.2 can't handle it properly. I've had to start nexus with:
-Des.cluster.routing.allocation.disk.threshold_enabled=false
to work around the issue.
Since elasticsearch is running in one node configuration perhaps it should be a default setting anyway (es node won't be able to relocate any shards ever).