we have ES cluster 6.4.2 with 2 nodes each node HW is:
Intel® Xeon® E3-1275 v5 Quad Core
64 GB DDR4 ECC RAM
2 x 512 GB NVMe Gen3 x4 SSD (Software RAID 1)
1 Gbit eth between them.
We have 1 index with 5 shards and 1 replica (tried different setups, but it was worse or equally bad). According to Search Profiler in Kibana the query would normally took about 10ms to get the result. What it really took is almost 300ms, because the 290ms is spent on rewriting the queries. Can the rewrite_time be somehow optimised or turned off or what can we do?
Thank you very much for you kind response.