while taking a look at the existing additional facet implementations
floating around, some questions raised to me:
Is a FacetProcessor instantiated per shard? As FacetProcessor.parse()
uses a SearchContext as parameter, I cant think of another case.
If the above statement is true, then is there any last sort of reduce
step for the per-shard results? If so, where does it happen? I havent found
it in a quick peek...
If none of the above is true I am open for explanations what happens and
how
You are correct, FacetProcessors are created per shard to generate an initial facet for the data on the shard. Tose facets are sent to the originating node ( the one that got the request from the client) where they are combined. This is also the responsibility of the FacetProcessor.
On Saturday, December 8, 2012 11:11:28 PM UTC+1, Alexander Reelsen wrote:
Hey there,
while taking a look at the existing additional facet implementations floating around, some questions raised to me:
Is a FacetProcessor instantiated per shard? As FacetProcessor.parse() uses a SearchContext as parameter, I cant think of another case.
If the above statement is true, then is there any last sort of reduce step for the per-shard results? If so, where does it happen? I havent found it in a quick peek...
If none of the above is true I am open for explanations what happens and how
I used your presentation as a reference already (thx for that one!)
Just hadnt seen, that the FacetProcessor from the querying node is also
responsible for calling reduce().. Will check the source for that.
Thanks for the clarification and have a nice weekend!
You are correct, FacetProcessors are created per shard to generate an
initial facet for the data on the shard. Tose facets are sent to the
originating node ( the one that got the request from the client) where they
are combined. This is also the responsibility of the FacetProcessor.
On Saturday, December 8, 2012 11:11:28 PM UTC+1, Alexander Reelsen wrote:
Hey there,
while taking a look at the existing additional facet implementations
floating around, some questions raised to me:
Is a FacetProcessor instantiated per shard? As FacetProcessor.parse()
uses a SearchContext as parameter, I cant think of another case.
If the above statement is true, then is there any last sort of reduce
step for the per-shard results? If so, where does it happen? I havent found
it in a quick peek...
If none of the above is true I am open for explanations what happens and
how
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.