freezing system state
glindahl at hpti.com
Fri Sep 15 09:44:37 PDT 2000
> Maybe what you are really looking for is a way to stop one or more
> processes and save them to disk, instead of the entire system state. This
> is a feature I would think could be extremely useful, but I'm not sure
> anyone is implementing a way to do this. Anyone know?
There are several such systems. Condor does it, RWCP's Score-on-Linux does
it, and there's a guy in Brazil who has a fairly ugly x86-only checkpoint.
You often have to add hacks to support stuff, i.e. re-open MPI sockets.
More information about the Beowulf