: It seems that TermComponent is looking at all versions of documents in the index. : : Does this is the expected behavior for TermComponent? Any suggestion about how to solve this?
Yes... http://wiki.apache.org/solr/TermsComponent "The doc frequencies returned are the number of documents that match the term, including any documents that have been marked for deletion but not yet removed from the index." If you delete/replace a document in the index, it still contributes to the doc freq for that term until the "deletion" is expunged (either because of a natural segment merge, or forced merging due to optimize) The reason TermsComponent is so fast, is because it only looks at the raw terms, if you want to "fix" the counts to represent visible documents, you have to use something like faceting, which will be slower becuase it checks the actual (live) document counts. -Hoss
