I have a cluster with 2 data nodes (5 shards on each). When looking in marvel I see that it says we have ~350million documents and 150million deleted documents, so we were thinking of running an optimize command (we dont really ever run them) in order to remove these deleted documents and help search performance.
Reading through the forums it seems like maybe its not recommended/needed to run a manual optimize. Although Seeing that close to half of the documents are deleted, I would think it would be a good thing to remove them all.
Are optimizes 'cluster' wide or if we do this would we issue an optimize command on each of the nodes?
is there a specific amount of space needed to run an optimize? I remember when working with solr that we needed at least 2x the size of the index for an optimize to run, is this true for elasticsearch?