HPC and Data for Lattice QCD
caosd
Using 'caosd'
caosd is a simple program which keeps control of concurrent
requests for using parts of a APEmille machine via the caos
operating system.
caosd runs on the apemaster and listens on port 10999 for
messages from caos. If the environment variable USECAOSD
is set to '1', caos will contact the caosd and try to book
the requested resources. caosd maintains an internal list
of currently running jobs. In order to avoid a wait-for-ever
situation it checks whether those caos processes which booked
the machine are still running.
There are some further administration utilities to control
caosd, which are not yet in cvs: bookmanager (list booked jobs),
book (book machine by hand), free (free machine by hand).