Just wondering if someone can point me in the right direction in relation to a problem I have been grappling with for about 12 months. We have a SharePoint 2010 system. We have set up numerous team sites along with 10 record centre sites. On a weekly basis (weekend) we transfer documents that have not been modified in 2 years to one of the 10 record centres using site collection information policies. The whole process works well except for the fact that for about 12000 documents in a system of almost 2 million total records, duplicates are created each week in Record Centre. In an effort to trouble shoot the problem I disabled all information policy, expiration and content organiser timer jobs in central admin to stop the weekly transfer of records. To my amazement, although the record transfer did not occur that week, i.e. no new documents were created in record centre, those documents that were known to duplicate in the past continued to duplicate. Interestingly the duplicates get created at approximately the sametime they would have been created had I not disabled the timer jobs. I repeated this several times over the last 12 months with the exact same results. The transfer doesn't occur but the same 12000 or so duplicates get created without fail.
I ran an SQL script against the AuditData table for one of the record Centre databases in question and noted that duplicates that were created appeared to be the result of a record submission: the EventData column specified that the records had been created as the result of a Records Respository Submission. What is puzzling is that the documents that apparently were submitted as duplicates no longer exist in the team sites they supposedly came from (I checked AllDocs, AllDocStreams and the RecylingBin to make sure none of them existed) and have not existed for about 5 years. This raises the question where are the documents coming from? The only thing I can think of is that somehow the event receivers in the record Centre libraries are playing up or have become corrupted. But if that is the case what triggers them to misbehave at almost the exact time a normal transfer would have occurred regardless of whether the timer jobs are disabled or enabled.
I restored one of the record centre databases to another environment to see if I could replicate the problem there but the behaviour did not repeat. This would suggest the event receivers might not be the issue. I was wondering if anyone has come across this kind of behaviour before or has suggestions on where I should look next as I have completely run out of ideas.