need a intelligent & distributed cron scheduler

I am having a process named replicate which runs on each machine/host a long time and consumes a lot of resources (disk-io, network-bandwidth, high CPU)
Goal is to reduce the network load by running all process replicate overall all hosts in a sequence
The replicate process is highly dependent on the previous data of the host and is therefore changing its runtime/exectime

I need a distributed cron job scheduler, which is able to:
- register new hosts, which need to be distributed scheduled
- to monitor start and end time
- to monitor missing starts or missing end times (operation not started or operation not terminated)
- to manage all replicate processes from a central point
- to define an order of the hosts
- to define group of hosts, which have to run the hosts of the given group in a sequence (e.g. datacenter US-CA, datacenter US-NY, ... to limit bandwidth per datacenter), but the groups itself can run in parallel (e.g. host in sequence of US-CA and hosts in sequence of US-NY can run in parallel)
- the hosts fetch the distributed cron and update the local cron accordingly

Programming languages allowed?
- bash
- java

delivery:
- into our git (will be shared to you)
- well documented code (inline and in form of a preview as README.md)