Cron downtime regardless of setup?

[latisaron]

Hi there, I have been met with a situation where using when resque-scheduler, downtime is unavoidable.

To give more context:

1. resque-scheduler-version: 4.7.0 (don't judge me please)
2. resque.rake:

# boilerplate
Resque.schedule = YAML.load_file('config/dummy_schedule.yml')
# boilerplate

3. dummy_schedule.yml

random_job:
  cron: "* * * * *"
  queue: generic
  custom_job_class: "ResqueRandomJob"
  description: "Runs every minute

On a multi-scheduler setup, where I reach a situation when I have:

• 1 master
• 1 child

If the master somehow dies, there are a few seconds (based on the lock-acquiring time) where the jobs do not run at all. I tested this by killing the process kill -9 pid at XX:XX:58, the next run being at XX:XX+1:00. This means that the child must become master, but in that situation, it might take a variable amount of time, based on where in the interval the child is, to become master again.

Therefore, in that one specific minute, there will be a missed enqueued job.

I was wondering if I'm doing anything wrong in this setup;

I also have an idea for a solution, but first I would like if I'm making any mistakes in this setup.

Thank you!