Our CPU node check is passing again: CPU nodes create, stop, and wake normally with their files intact.

Our continuous check of CPU nodes - creating a real cpu-2 node, stopping it, and waking it with its files intact - is failing, and we are investigating what it caught. Creating a new CPU node, or waking a stopped one, may be delayed or may fail while this is open. GPU nodes run on separate machines and are not affected, running CPU nodes keep running, and stopped CPU nodes' files remain safe in durable storage. If a create or wake fails, nothing is lost or charged: it is safe to try again once this is resolved.