A nightly job finishes in two seconds and is not available when Prometheus next scrapes. The job records a service-level completion timestamp. Which approach fits this narrowly scoped use case?

1. Run a Pushgateway on every long-lived web server as the default collection path
2. Set the scrape interval to one hour
3. Configure a recording rule to reconstruct missing scrapes
4. Publish its service-level metric through a Pushgateway and manage the metric lifecycle