Purging old data
The built-in implementation of the persistence API for relational databases never deletes saved data from your database. When users "delete" data from the UI, it marks the data as deleted in the database, and then ignores such data. This is done to increase safety, enable auditing, and allow an admin to "undelete" data that has been deleted by mistake.
While there are benefits to keeping data marked as deleted, there are cases where you might want to get rid of it, for instance to comply with regulatory requirements, or for larger systems to save space and improve performance.
This is done in 2 steps, starting by removing form data, and then the corresponding attachments. We recommend you follow the steps below only if you are familiar with SQL. Also, especially if you are going to run this on a production database, we highly recommend you first create a backup of that database.
The following queries purge data that users manually deleted. However, should you want to adapt those queries to also purge "historical data", keep in mind that forms you edit in Form Builder are also stored in the
orbeon_form_data_attachtables. This means that in your own version of those queries, you might want to avoid rows with with
app = 'orbeon' AND form = 'builder', so not to delete any Form Builder data and so not to loose the ability to open forms from the Form Builder summary page.
The following statement is for Oracle (if you're using another database, replace
ADD_MONTHS(SYSDATE, -1)by an equivalent function on that database). It will return the data marked for deletion more than 1 month ago. If you want to delete that data, replace
DELETEin the statement, and run it again.
WHERE document_id IN
FROM orbeon_form_data t,
SELECT max(last_modified_time) last_modified_time,
app, form, document_id
GROUP BY app, form, document_id
WHERE -- Look at "last row" in the "journal"
t.last_modified_time = m.last_modified_time AND
t.app = m.app AND
t.form = m.form AND
t.document_id = m.document_id AND
-- Take deleted items
deleted = 'Y' AND
-- Deleted at least 1 month ago
t.last_modified_time <= ADD_MONTHS(SYSDATE, -1)
Next, we want to remove "orphan" attachments. Those are attachments no longer referenced by an existing document. As in the previous step, first check the data returned by the following query, and make sure this is data you want to delete. If that is the case, replace
DELETEand run that statement again.
WHERE document_id NOT IN
orbeon_i_control_texttables may contains references, via the
data_idcolumn, to entries in
orbeon_form_data. The content of these tables is recreated when a full database reindex is performed.
If the lease feature is enabled and in use, the
orbeon_form_data_leasetable may also contains references, via the
document_idcolumn, to entries in
If that is the case, the relevant rows must also be deleted.