I have been experimenting with using Deduplicate for CLOBs, but have repeatedly hit a deadlock situation.
Then I tried creating a very simple test case where one session inserts a large CLOB string into a table and sleeps before commiting, while a second session inserts a separate row but with the exact same value for the CLOB string. This second session will block until the first one is commited even if it's an independent row.
When you have a number of threads inserting batches of values into a table with Deduplicate option on, this is very likely to lead to deadlocks if the same values are common.
Is this working as intended? This seems to severely limit the use cases for which the deduplicate option is appropriate.