I have 3 unique servers for elasticsearch, i made the cluster setup in this.
Each and every index has 5 shards and 2 replicas. i have an index with 4
lac of documents in it. while perform a search query in the elasticsearch
cluster. the returned documents have different calculated score in
different servers.
ex
Document Server 1 Server 2 Server 3 ----------------------------------------------------- documentA 49.379543 50.168644 49.577785 documentB 17.744095 17.979729 18.046381 documentC 16.880697 16.748943 16.850113 documentD 10.785672 10.744943 10.806508
.
.
search_type*=dfs_query_then_fetch*
This is because of the document frequency. but i can;t understand that all
are calculated in shard0 itself. if we set the search preference to
_primary then the score same for all the request (means all the servers).
obviously this should be. Because there is no other search hits for the
replicas.
My Question is: what's happen if the search hits the replicas? it
calculates different score in different servers?
On Friday, 6 March 2015 19:27:07 UTC+5:30, Shanmugam wrote:
Hi All
I have 3 unique servers for elasticsearch, i made the cluster setup in
this.
Each and every index has 5 shards and 2 replicas. i have an index with 4
lac of documents in it. while perform a search query in the elasticsearch
cluster. the returned documents have different calculated score in
different servers.
ex
Document Server 1 Server 2 Server 3 ----------------------------------------------------- documentA 49.379543 50.168644 49.577785 documentB 17.744095 17.979729 18.046381 documentC 16.880697 16.748943 16.850113 documentD 10.785672 10.744943 10.806508
.
.
search_type*=dfs_query_then_fetch*
This is because of the document frequency. but i can;t understand that all
are calculated in shard0 itself. if we set the search preference to
_primary then the score same for all the request (means all the servers).
obviously this should be. Because there is no other search hits for the
replicas.
My Question is: what's happen if the search hits the replicas? it
calculates different score in different servers?
On Friday, 6 March 2015 19:27:07 UTC+5:30, Shanmugam wrote:
Hi All
I have 3 unique servers for elasticsearch, i made the cluster setup in
this.
Each and every index has 5 shards and 2 replicas. i have an index with 4
lac of documents in it. while perform a search query in the elasticsearch
cluster. the returned documents have different calculated score in
different servers.
ex
Document Server 1 Server 2 Server 3 ----------------------------------------------------- documentA 49.379543 50.168644 49.577785 documentB 17.744095 17.979729 18.046381 documentC 16.880697 16.748943 16.850113 documentD 10.785672 10.744943 10.806508
.
.
search_type*=dfs_query_then_fetch*
This is because of the document frequency. but i can;t understand that
all are calculated in shard0 itself. if we set the search preference to
_primary then the score same for all the request (means all the servers).
obviously this should be. Because there is no other search hits for the
replicas.
My Question is: what's happen if the search hits the replicas? it
calculates different score in different servers?
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.