Hello, I'm making a BOINC project with the pDynamo program (molecular modeling) https://sites.google.com/site/pdynamomodeling/ . It is a big library of Python and C. It runs via Python scripts (input.py), and writes an output.

I read on the wiki that I can use pyBOINC, but PyBOINC is not maintained. Can I use vboxwrapper, with a virtual machine with the library compiled inside?

I can't use py2exe since i will run on unix guests and it is a library made of python, c and cython.

I already have the server up and running!

Are there any other options? Will the vboxwrapper work?

Thank you.
J

The only one I know about would be vboxwrapper

In theory you could create a python virtualenv for each OS and copy that, but it'd be rather difficult (perhaps a nightmare) to administer

---

Regards
Kevin
-------------
theSkyNet POGS
http://pogs.theskynet.org/pogs

The vboxwrapper approach would be the best in this case. You should create your VM with all the libraries and executables installed. Make sure this is not too big and give it a try. How long are your computations? The vboxwrapper is currently not so friendly with longer computation times. But this is going to be fixed in the near future.

Hi, thanks for the feedback.

My computations will not take more than 2hours/user, is it ok for the vbox?

I will now configure the VM!

And another question!

Since its a library I don't really have a "boinc_app" to monitor. I will just call the input scripts. The input would be the "boinc_app".

I'm thinking on doing a really simple program that just calls the input file via command line and waits till a output file is written. Is this the best solution? Is it possible to just copy the input file as a "boinc_app"?

The "boinc_app" can be a script of course. It just has to be executable. At RNAWorld we have a batch script that calls the real executable, monitors the memory usage and copies the output files into the right place so the client can find them.

You should also implement something like this if only to use it later. You have to make sure that the output files are in the shared/ folder and have the correct name that is specified in the output template.

If you are calling scripts make sure not to execute them inside the shared/ folder as this will not work with snapshots.