We observe issues when starting all jobs together because this can lead to a race condition with multiple writes to kube config file, since all clusters will be finished at roughly the same time.
Thus, we introduce a sleep after each job so clusters won't all be created at the same time.
/assign @boredabdel