When does a database checkpoint occur?

When an online redo log group becomes full, the database switches to the next available group, which typically initiates a checkpoint. This process ensures that all modified data blocks associated with the log group being switched out are written from the database buffer cache to the data files on disk, synchronizing the control file and data file headers with the latest changes.

Abnormal termination of a user session triggers a rollback of uncommitted transactions for that specific session, not a database-wide checkpoint.

Abnormal termination of a server process might lead to process recovery but does not directly cause a database checkpoint.

Issuing the SHUTDOWN ABORT command immediately terminates the database instance without performing a checkpoint or a clean shutdown, necessitating instance recovery upon the next startup.