K8S Version: 1.23

I have an hourly CronJob with 5 inner pods. After some time, part of these pods goes to shutdown and moved to the Completed state (it's expected), but the rest - still works, so have a Running state.

At the next hour, this CronJob will not be triggered due to the Running pods (it's also expected). But I need to force recreate pods with the Completed state if a Running still exists. Is it possible?

CodePudding user response：

It appears that the reason you're letting the Running pods exist is that you expect those pods to take a long time before finishing. This means that the scheduling of those pods should be different from others, since they can take a longer time to finish.

You can split your CronJob into two CronJobs. One of these will run every hour and will only have the pods that get Completed. The other will run less frequently (maybe every 2 hours?) allowing the Running pods to finish.

This way, you will be able to manage your cron tasks separately.

Note: For k8s version, we usually mention the API version which is of the form v1.xx. It appears you are reporting the version of a public cloud offering of k8s like AKS, EKS or GKE. Just FYI.