SOLR-5057 introduced a bug in queryResultCaching such that the following circumstances can result in a false cache hit...
- identical main query in both requests
- identical number of filter queries in both requests
- filter query from one request exists multiple times in other request
- sum of hashCodes for all filter queries is equal in both request
Details of how this problem was initially uncovered listed below...
uwe's jenkins found this in java8...
The seed fails consistently for me on trunk using java7, and on 4x using both java7 and java6 - details to follow in comment.