I have several systems/laptops which are dual boot between Linux and Windows. I also partition the disks such that I can share/access a FAT32 partition from either system. My question is - Is it possible to set up the client software such that I am effectively using the same data/projects on each system? What I want to be able to do is to switch between the OS's and have BOINC and all my projects keep running from where they left off, regardless of which OS was used last. If this is doable, pointers on how to set up the file system and any underlying files would be appreciated.

I will go poke at it myself, but just thought I might save myself some time by asking first, in case anyone has done it already.

Marc..

---

Marc Chamberlin

Computers: the final frontier. These are the voyages of the user Marc.
His mission: to explore strange new hardware. To seek out new software and new applications.
To boldly go where no Marc has gone before!

No, that isn't possible, due to the fact that any work you download from a project is registered to a certain hostID. That same hostID should also upload & report the work.

Since we're talking about two different OSes, you'll get two hostIDs and they are seen as different computers, even if they live on the same system.

Sharing the data directory can probably be done (although a permission nightmare), but running the same work on both OSes cannot.

---

Thanks Jord, that's too bad, I keep losing a lot of work because of my switching OS's and I tend to use each for a length of time that exceeds the project expiration(s) on the other OS... Sigh, Oh well..

Marc..

---

Marc Chamberlin

Computers: the final frontier. These are the voyages of the user Marc.
His mission: to explore strange new hardware. To seek out new software and new applications.
To boldly go where no Marc has gone before!

Excuse me for butting in here but yes this can be done and here's how. I'm not going to suggest a perfect solution here but I'll explain why it's not perfect. First install Virtualbox on every OS that is installed natively and also remove the boinc client from them. Next using Virtualbox create a virtual machine and put it's virtual hard disk on your data partition. The virtual hard disk doesn't have to be huge 10 to 20 gig should do it. Give the virtual machine access to all of your system CPUs and almost all of your system ram. Install a lightweight operating system on this virtual machine, the smaller the better, I suggest DSL or Puppy Linux. Now install boinc on that virtual machine and connect virtualbox on all of your native operating systems to that same virtual machine. Whoala all of your native operating systems are now processing the same boinc tasks. The reason this is not a perfect solution is 1 bionc will not run unless the virtual machine is running so you will have to remember to stop and restart the virtual machine every time you reboot the physical machine. 2 virtualization incurs overhead so the tasks won't complete as quickly as they would if bionc was running on the native OS. Expect it to be about 3% slower. 3 Virtual machines can't directly access your video card so it will not be able to process GPU tasks. If you're not processing GPU tasks now then this obviously is not an issue.

No, that isn't possible, due to the fact that any work you download from a project is registered to a certain hostID. That same hostID should also upload & report the work.

Since we're talking about two different OSes, you'll get two hostIDs and they are seen as different computers, even if they live on the same system.

Sharing the data directory can probably be done (although a permission nightmare), but running the same work on both OSes cannot.

Registration to a HostID, that too but unless you do pure math [Though I've seen claims that taking a data_dir set to another machine, completing them and reporting them from that other machine works], mostly Bioscience/chemical apps for Linux don't process exactly the same as for Windows or Mac for that matter... that's why e.g. WCG maintains homogeneity groups so Linux results can be matched with other Linux results any time a quorum or a repair is required.

Dual boot between different Windows versions did work, sharing the same data_dir, but who's doing that these days?

---

Coelum Non Animum Mutant, Qui Trans Mare Currunt

I have started a climateprediction model on one computer then transferred it to another computer and finished and reported it from there. So for CPDN the different computer ID doesn't seem to worry the server. But both computers must have the same type of OS because the model compilation is different for each.