When a run's checkpoints and events become durable. In every mode a durable event is delivered only after the commit that numbered it, so no subscriber sees a sequence number that a crash could later reuse. Suspensions (#1269) will be persisted synchronously in every mode before a suspended result is returned.
Commits are queued and written in order on a background writer while the run continues. After a crash, the work since the last durable checkpoint is redone by recover. The run waits for the queue before returning, and fails if any queued commit failed.
Commits are queued and written in order on a background writer while the run continues. After a crash, the work since the last durable checkpoint is redone by recover. The run waits for the queue before returning, and fails if any queued commit failed.
Nothing is written until the run ends; then the last checkpoint, its pending writes and every buffered event are committed at once. A crash loses the whole run's progress.
Nothing is written until the run ends; then the last checkpoint, its pending writes and every buffered event are committed at once. A crash loses the whole run's progress.