Looking at icinga UNKNOWN errors, It shows that shard size check and unassigned shards check are failing with
UNKNOWN - HTTPConnectionPool(host='localhost', port=9200): Read timed out. (read timeout=4)
However, testing manually on the one of the nodes that reported failure with an increased timeout seems to not throw this error.
We should figure out why this is happening as resources in cloudelastic hosts does not show a very high load.