Hey!

I am currently in the process of researching the possibility of running multiple machines that run BOINC computations and only having one central computer that is interfacing with BOINC servers. For this I assume I would have to modify the client so that the logic for execution is separate from the one managing tasks.

Do you know how this can be done and if this is even possible?

Also, if you could point me to someone knowledgable in BOINC code that would be willing to discuss this on a call, that would be great.

Any help would be much appreciated.

The idea is called a 'SuperHost', and as this document says, has been proposed many times before.

https://boinc.berkeley.edu/trac/wiki/SuperHost

"not yet been implemented" it says, and boy, is that still true - even though that wording was in the very first version of the document, 13 years ago.

Best of luck - we'd all like to see this put into effect.

I am currently in the process of researching the possibility of running multiple machines that run BOINC computations and only having one central computer that is interfacing with BOINC servers. For this I assume I would have to modify the client so that the logic for execution is separate from the one managing tasks.

It does support use of a proxy server if you want a separate machine to communicate with the outside world, but it sounds like you were after more than that.

---

MarkJ

In an ideal world I would like to have the execution layer of BOINC separate from the client, so that it can be moved to a different machine or delegated to someone else. Maybe this can be somehow done without going 100% the Superhost way, just some dirty hack making the separation possible in the code.