The current values for
cluster.routing.allocation.disk.watermark.low cluster.routing.allocation.disk.watermark.high
are too low -- 70% and 85%. On a 4 TB disk this means we can leave 1.2 TB unallocated.
We should decide on and implement new, higher defaults. I'll propose low=85% and high=90%.
We should also update the docs as part of this work as they reference the defaults.