My current cluster is running on ES 1.7.3 with several nodes. I use only 2 shards for index.
Every week I move the 2 shards of my index to another nodes and the search query_time_in_millis decrease too much p.e.: 800ms to 40ms.
Why this stats decrease so much after relocate the 2 shards?
I'm not sure if this stat is trustly, my java application who use this Elasticsearch cluster has not logs to measure this byh now.
Can I use this metric to measure my application performance?
For other side, I try to look for the documentation about this measure descrition (search - query_time_in_millis in INDEX api) but today i'm not a lucky man.
Could you share me the documentation link?