[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [HTCondor-users] Jobs on Windows and heterogeneous pool

On Thu, May 16, 2013 at 11:33:09AM +0000, Romain wrote:
> Targets are to can submit and run jobs from and on all machines.

I'm not an expert here, but I think life is easier if you submit/run jobs
directly on one machine.

Otherwise, as I understand it:

- it's possible to have a single schedd instance, but you need to make
  some config changes or supply some extra options to condor_submit /
  condor_q / condor_status etc to use a remote schedd.

- I think everyone could run their own schedd, which means they all have
  their own private job queues.  I'm not sure of the benefits and costs of
  doing it that way.

> Wine is to run Windows applications on Linux no? So I would prefer run Linux 
> applications on Windows to do that I have to use cygwin ?
> It's something like virtual machine which link Windows and Linux "emulation"
> (cygwin) it's that ?

Not exactly.

Cygwin gives you a posix API layer. It lets you *compile* a version of the
same source code to run under Windows.  But you can't run the same binary on
both Linux and Windows.

But you can build app.Linux and app.Windows binaries, and condor can select
the right one to run on a particular host.

WINE attempts to be a full binary compatibility layer for running unmodified
Windows apps under Linux. It achieves variable levels of success.

> If I use cygwin can I install all my applications, and Condor like on a 
> Linux machine ?

In principle you might be able to build Htcondor from source under Windows,
but Windows sucks so badly I wouldn't want to try it.  I suggest you use the
pre-built Htcondor Windows binaries.

However Cygwin will give you bash, perl, awk, sed, ...  so that it's easier
to write scripts which run on both platforms.

Anyway, as I avoid Windows completely I'm afraid I can't really help you
much more.