what should be the relation between search response "took" value and
incremental change reported
by NodeStats.indices.search.query_time_in_millis ? Should they be in sync?
I am testing it with 0.19.0 and I can see they can differ sometimes
significantly (for example, search took = 75ms, incremental change in
query_time_in_millis = 156ms, or 36ms vs 19ms).
Sometimes I even notices situation, where query was executed but both
response "took" and NodeStats.indices.search.query_time_in_millis reported
0ms
In my test I run string query with several facets and highlights. A single
node.
The "took" part of a search response refers to the whole search execution
API process. The query_time part in node stats or indices stats just refers
to the time the query executed on a specific shard (or the shards on that
node).
what should be the relation between search response "took" value and
incremental change reported
by NodeStats.indices.search.query_time_in_millis ? Should they be in sync?
I am testing it with 0.19.0 and I can see they can differ sometimes
significantly (for example, search took = 75ms, incremental change in
query_time_in_millis = 156ms, or 36ms vs 19ms).
Sometimes I even notices situation, where query was executed but both
response "took" and NodeStats.indices.search.query_time_in_millis reported
0ms
In my test I run string query with several facets and highlights. A single
node.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.