Distributed Result Grouping Caveats
Grouping is supported distributed searches, with some caveats:
1)Currently group.func is not supported in any distributed searchers.
2)group.ngroups and group.facet require that all documents in each group must be co-located on the same shard in order for accurate counts to be returned.
Document routing via composite keys can be a useful solution in many situation.
Shards and Indexing Data in SolrCloud
Document Routing
1)compositeId
two levels of routing; shard_key/num!document_id;
2)implicit
ps.
Ingoring Commits from Client Applications in SolrCloud
openSearcher=false/true
softCommit/hardCommit