Failed Publishes Eating Up Space

G'day,

We attempted an initial publish of about 10K assets, a mix of content only, images, and videos of a site. Owing to infrastructure issues, the publishes kept failing.

As a result, we hit the limits of the user tablespace. Each time the publish failed, another 5 GB (for the data we had) was consumed.

This we traced to the FW_PubDataStore which stores the binary content in a lob segment. So even though the main table remains small, the data actually occupies a lot more space.

If you delete the publishing history, the space is not reclaimed. You need to drop into Oracle to truncate the tables. Is this expected? What is the best practice for dealing with this? To use a scheduled job to compact the table?