Description
The reduce() method of SolrDeleteDuplicates deduplicates SOLRRecords given their signature. The first SOLRRecord is stored in a variable recordToKeep and is compared to the following SOLRRecords found with the same signature. The only trouble being that the first instance is reused by Hadoop when calling values.next() and hence recordToKeep gets the same values as the latest call to values.next().
The patch attached clones the SOLRRecord before assigning them to recordToKeep in order to avoid the problem.