CODEWITHSUNDEEP
kubernetescronkubernetes-cronjob
David S.
David S.

Reputation: 6105

How to control interval between kubernetes cronjobs (i.e. cooldown period after completion)

If I set up a kubernetes cronjob with for example

spec:
  schedule: "*/5 * * * *"
  concurrencyPolicy: Forbid

then it will create a job every 5 minutes.

However if the job takes e.g. 4 minutes, then it will create another job 1 minute after the previous job completed.

Is there a way to make it create a job every 5 minutes after the previous job finished?

You might say; just make the schedule */9 * * * * to account for the 4 minutes the job takes, but the job might not be predictable like that.

Upvotes: 0

Views: 460

Answers (1)

Dawid Kruk
Dawid Kruk

Reputation: 9877

Unfortunately there is no possibility within Kubernetes CronJob to specify a situation when the timer starts (for example 5 minutes) after a job is completed.

A word about cron:

The software utility cron is a time-based job scheduler in Unix-like computer operating systems. Users that set up and maintain software environments use cron to schedule jobs (commands or shell scripts) to run periodically at fixed times, dates, or intervals.

-- Wikipedia.org: Cron

The behavior of your CronJob within Kubernetes environment can be modified by:

  • As said Schedule in spec definition 
      schedule: "*/5 * * * *"
  • startingDeadline field that is optional and it describe a deadline in seconds for starting a job. If it doesn't start in that time period it will be counted as failed. After a 100 missed schedules it will no longer be scheduled.
  • Concurrency policy that will specify how concurrent executions of the same Job are going to be handled:
    • Allow - concurrency will be allowed
    • Forbid - if previous Job wasn't finished the new one will be skipped
    • Replace - current Job will be replaced with a new one
  • Suspend parameter if it is set to true, all subsequent executions are suspended. This setting does not apply to already started executions.

You could refer to official documentation: CronJobs

As it's unknown what type of Job you want to run you could try to:

  • Write a shell script in type of: 
    while true
    do
      HERE_RUN_YOUR_JOB_AND_WAIT_FOR_COMPLETION.sh
      sleep 300 # ( 5 * 60 seconds ) 
    done
  • Create an image that mimics usage of above script and use it as pod in Kubernetes.
  • Try to get logs from this pod if it's necessary as described here

Another way would be to create a pod that could connect to Kubernetes API.

Take a look on additional resources about Jobs:

Please let me know if you have any questions to that.

Upvotes: 1

Related Questions