I've got a dev cluster running elasticsearch 1.7.0 that I haven't had need to look at very closely for a couple months but now I'm prepping to to a rolling replacement of the instances and a noticed an oddity. When I first set it up everything was looked fine but now I notice that most indexes have all their primary shards on the same node. A few have them split a bit but even those have most of the shards on that node.
So I have 3 questions which I'd appreciate any guidance on:
- Is this a problem?
- Do I have settings screwed up somewhere (see below)?
- Does this need to be fixed before I do the replacement of all my nodes?
Thanks in advance