When there is an exception occurred while the job was running, like network shake or ZK service shake. The job will be retried. The configuration "kylin.job.retry" will handle this.
But sometimes we need to pause a little time before the retry job start. The purpose is to wait for the network shake or ZK shake recovered, otherwise the retry make no sense.
The design is to add a configuration "kylin.job.retry-interval", the default is 30000 milliseconds.