Long-running tasks are being prepared for suspension and resumption using a dedicated compute environment. This allows workloads to be snapshotted mid-execution and spun back up later, which should free up resources during idle periods.
Virtual machine creation, snapshotting, and restoration are routed through a central gateway API by a new . When a task needs to be suspended, a is triggered, and an asynchronous callback confirms when the state is safely stored. Upon resuming, the exact CPU, memory, and environment metadata are injected back into the restored machine before it wakes up.
The feature is activated automatically when a is configured in the environment.