To: mbarker@MIT.EDU Cc: mbarker@MIT.EDU Subject: long jobs Date: Wed, 11 Dec 1996 00:38:31 EST From: Mike Barker suppose we assumed 1 login - kerberos authenticated if you want to run a job longer than normal tickets allow, it is up to you to either login regularly and renew or do your work uslng local disk. i.e. one cpu, one user-very simple! forget the whole multiprocessing, batch job paradigm and use old cpu's. we might have a simple reservation scheme-a web page list and zephvr notification when a cpu was free?