Cron jobs about kured

[awh]

Hi @vitobotta - interesting question. Are these cron jobs configured at the OS level or are they kubernetes CronJob resources? If they're the latter, you can use the new blocking reboots via pods feature in kured 1.2.0. If it's the former, I'm not sure what will happen to be honest - kured just runs systemctl reboot, so unless that happens to interact gracefully with running cron jobs and lets them run to completion, they'll probably just get terminated.

[vitobotta]

Hi @awh , I was referring to Kubernetes cron jobs since I prefer not doing anything outside K8S if I am using K8S for my apps. I checked that page but there would be a problem with the name of the pods used by cron jobs because they change. For example when MySQL jobs are running the jobs are named something like mysql-cluster-auto<timestamp>-backup. Any ideas of how to handle this? Thanks!

[awh]

@vitobotta great, that makes things much simpler. You can specify additional labels on the pods created by CronJob resources - you have to get them in the right place though:

apiVersion: batch/v1beta1
kind: CronJob
metadata:
    name: backup-cronjob
spec: # CronJobSpec
    jobTemplate: #JobTemplateSpec
        spec: #JobSpec
            template: # PodTemplateSpec
                metadata:
                    labels:
                        block-reboots: true
                spec:
                        ...

then add

--blocking-pod-selector=block-reboots=true

to your kured configuration.

[vitobotta]

I will have to check if I can change the specs of the cron jobs but it seems a little complicated because they are created by the MySQL operator. I was actually mistaken in that they are not exactly cron jobs in Kubernetes terms, they are one-time jobs created by a separate scheduler which is in the operator, or at least I think it works this way more or less. I will have to investigate. Thanks for pointing me to the blocking option, I'm closing this one for now. :)