TODO.txt

--- TODO ---
scale proportionally (using a scaling factor) to the job request rate + queue size
make allocation message to workers, which allows for:
	- also add IP to this message so the worker can check whether the right client is providing the data
	- notifying the controller if the job data never reached the worker
	- checking if result request matches ip of job

split controller into multiple files
nicer experiment administration
nicer experiment framework at client

what if worker is killed / dies / scaled down before client has received all data
what if worker crashes while it is pending?
maintain precedence over restarted jobs + account for job restarts in the job completion time

retrieve status reports as client until result is produced (polling)
controller faster ip resolution
clean client interface + meerdere file uploads
remove debug crap
allow again for controller restarts

write bootstrapper for controller that hands it its ip

--- QUESTIONS/POSSIBILITIES ---
retrieve status reports as client until result is produced
priority system for jobs
increase security
less http

--- IF WE ARE RICH ---
Worker restarts after scaling down