rufus-scheduler — Ruby cron & interval scheduling

require 'rufus-scheduler'

scheduler = Rufus::Scheduler.new

# Cron
scheduler.cron '0 9 * * 1-5', tz: 'America/New_York' do
  Reports.send_morning_digest
end

# Interval
scheduler.every '5m' do
  Heartbeat.ping
end

# One-shot
scheduler.in '30s' do
  Cleanup.run
end

scheduler.join

• Standard 5-field: 0 9 * * 1-5
• 6-field with seconds at the front: 30 0 9 * * 1-5
• Descriptors: '@hourly', '@daily', '@weekly', '@monthly', '@yearly'
• Last day: 0 0 L * * (Fugit supports L)
• Aliases: 0 9 * * MON-FRI

# Daily 09:00 in New York, regardless of DST
scheduler.cron '0 9 * * *', tz: 'America/New_York' do
  daily_job
end

# Daily 09:00 UTC
scheduler.cron '0 9 * * *', tz: 'UTC' do
  daily_utc_job
end

# Without tz: uses the host system's timezone
scheduler.cron '0 9 * * *' do
  ambiguous_job
end

Always pass tz: in production. The host timezone may differ across deployments and changes silently with the OS configuration.

• In-process only — the schedule dies with the Ruby process. For durability, use sidekiq-cron or a system cron.
• Multi-process Rails — every Puma worker runs the scheduler unless you gate on Process.pid or use a leader-election lib. Otherwise the cron fires N times per tick.
• Long-running blocks — by default jobs run in their own thread, so a slow job doesn't block others. But many slow jobs at once can exhaust the thread pool.
• Overlap — pass overlap: false to skip a tick if the previous run is still going.