Goal:

Foreman has various recurring tasks that today run via cron. Foremanctl should avoid cron and instead try to provide container native solutions where possible.

Foreman plugins that depend on foreman-tasks should use recurring logic because it plays nicely into complex multi-server deployments. For plugins that don't or foreman-tasks itself a simple mechanism should be provided. That way a sprawl of jobs is avoided.

The framework for this is a set of rake tasks in Foreman core. It will provide them with their based on the schedule:

• foreman-rake cron:hourly
• foreman-rake cron:daily
• foreman-rake cron:weekly
• foreman-rake cron:monthly

Then foremanctl implements systemd timers to run these are the scheduled interval. The timers may use RandomizedDelaySec= or FixedRandomDelay= to avoid all jobs on the server running at exactly the same time, but is not required.

Background on where it was initially suggested: https://github.com/theforeman/foremanctl/pull/288#pullrequestreview-3472321405.

An upstream RFC should be created for this to describe the design. This allows pointing plugin developers to how it's preferred.

This limits users on selecting the exact schedule, but that's a feature. When upgrading we won't make an effort to detect the existing schedule. A release note should cover this change in behavior.

Acceptance criteria:

• In upstream an RFC is created that describes the design
• Foreman is enhanced to implement the hourly, daily, weekly and monthly rake tasks
• Existing cronjobs in Foreman and Satellite are either converted to foreman-tasks or the new scheduled jobs
• For upstream users that don't use a containerized installation (yet) the cronjobs deployed by packaging are updated
• Upstream a release note about this is written