I know how to sneakernet on Windoze. I experimented until I got it to work, (later found directions on some website). You need to import the PandeGroup registry key for the UserID. Otherwise it tries to connect to Stanford servers to get the ID. (Reg keys are different on Win98/ME and Win2K/XP/NT)
It should be easier with Linux, because, I think, it uses a file in the run directory to store the UserID.
You will have to use a Linux rig to get WUs to move to another Linux computer.
I use a USB thumbdrive to move the folder. I move the whole thing, but after the first time, you only have to move the Work folder and the queue.dat file, I think. I just always move everything because it's easier to keep track of what goes where.
You can configure to get deadlineless WUs and minimize networking, which will cache ten WUs and not need to connect until seven or eight are completed. Or you can do regular WUs and hope to get high point, long lasting WUs.
One thing you need to be careful of is MachineIDs. You can have only eight MachineIDs per UserID. F@H servers keep track of the WUs sent to each UserID / MachineID. So if you have two MachineID = 1, you will get the same WU to crunch twice, but only get credit once. The length of time the servers keep track of WU sent to you can vary.
Also, you cannot run two instances with the same MachineID on the same computer.
If you can't get it to work on Linux, search the Stanford forums. There is a Linux forum there.
