Apparently if you copy a document which was saved like 2 days ago to another database with CopyToDatabase, they modified datetime stays the same. So if you have a replication schedule on the target database of lets say every day, the document will not be replicated, because its too old.
What is even better (happened here, took me 2 days to find it):
Copy a document with copytodatabase to a target database in a cluster environment. Delete the copied document in the target database. The deletion stub is replicated immediately to the other cluster members. Now copy the same document again. It will get the same UNID in the target database again!!! But the modified date doesn't change, so the deletion stubs on the other cluster members are actually newer. Replicate and your document is gone.
Never ever use CopyToDatabase, it has bitten me so many times now. Create a new Document and use CopyAllItems. More info here:
{Link}
What is a real killer though:
{Link}
Apparently if you copy a document which was saved like 2 days ago to another database with CopyToDatabase, they modified datetime stays the same. So if you have a replication schedule on the target database of lets say every day, the document will not be replicated, because its too old.
What is even better (happened here, took me 2 days to find it):
Copy a document with copytodatabase to a target database in a cluster environment. Delete the copied document in the target database. The deletion stub is replicated immediately to the other cluster members. Now copy the same document again. It will get the same UNID in the target database again!!! But the modified date doesn't change, so the deletion stubs on the other cluster members are actually newer. Replicate and your document is gone.