Am facing an issue with counts when using group.facets in my sharded solr. (Groups do not overlap across shards and for various reasons I cannot use group.truncate)
Now, the problem being faced is that for items ranking lower in the faceted list sorted by count, the group facet counts are coming *higher* than actual values. So on doing an online search I came across details of sharded faceting at this link: http://lucene.472066.n3.nabble.com/At-a-high-level-how-does-faceting-in-SolrCloud-work-td4009897.html >From the above link it appears there is a *third corrective step* wherein the coordinator node after getting individual results and building a final list, asks each shard to compute it's exact count for selected constraints. I wanted to ask if the Group.facet implementation in 4.x has been factored in this step and that the coordinator node is asking for grouped facet values instead of ungrouped facet counts during the corrective step ? Asking this because, counts are coming right for the 50% of the popular items but are incorrect (and always higher) for lesser items. Also has anyone else faced this ? Ritesh -- View this message in context: http://lucene.472066.n3.nabble.com/Group-Facet-issue-in-Sharded-Solr-Setup-tp4116077.html Sent from the Solr - User mailing list archive at Nabble.com.
