Small sample of a heptio ark deployment in order to reproduce a possible bug of the scheduler. Core problem: the scheduling stops working if the deploy is executed again.
Tested with ark 0.10 (current stable version)
To reproduce:
- deploy with
make k8s-deploy S3_BUCKET=<MY-BUCKET> CLUSTER_NAME=<mycluster.k8s.mydomain-or-something.com>
- watch ark doing backups as intended (every 2 minutes according to the schedule yml)
- deploy again with the command from the first step
- schedule stops working, observable via Prometheus or
ark schedule get
in the "last backup"-column. In the logs the line "scheduler is due, submitting backup" is not appearing anymore.