Lock maintenance

You can create a script that allows you to identify locked records and unlock them automatically. A timeout script should perform these tasks:

• Collect the set of record types defined by the schema.
• Build a data structure that identifies all the locked records for each record type. The structure should include information such as the user that holds the lock, the version that is locked, and the time that the lock was observed. This structure is initially empty.
• Build a query to find records that are currently locked, selecting the record ID, locked_by and lock_version columns.
• Compare the current set of locked records with those already known and recorded in the data structure. This check must include a match on both the locking user and the version of the record that was locked, in order to distinguish locks by the same user but on a new version of the record or by a different user.
• Remove locks that have been held longer than a chosen interval, by loading the entity and calling the Unlock API method.
• Update the data structure to add references to records that are found to be newly locked and remove references to records that are no longer locked.

The script should be set to run at regular intervals. The chosen timeout interval and the frequency that the script performs a scan will determine the minimum and maximum expected time for a lock to timeout. These values should be made known to users, since they affect:

• how long a user can work on editing a record before losing their lock.
• how long a user must wait for an abandoned lock to be released automatically.