I don’t know much about running momentics on targets from an XP development environment but unless your targets are somewhat blind, no keyboard or terminal, why not just load them up and startup the programs manually for testing. In the end you have to find a way to do this anyway, right?
I might also add, its not a bad idea to start out running your whole system on a single node, unless it is specifically redundancy that you are after. Message passing is almost transparently the same whether over a network or not. The main difference from a process point of view is speed.
You can start more than one Run/Debug Job in the Eclipse API, referencing different client (Neutrino OS´). so both clients would run the given executables. You just then have to start both from the IDE.
Later on you would have need them to send a signal to a watchproccess or similar for starting up both automatically (or have them in an Running State since startup).
Yes, a single node is enough currently for me. But as a graduation project, I was required to use two computing nodes. And it is best for me to do all the operation on the host nodes.
To micro,
What you said is exactly all I want to implement, and I think your idea is really great. But could you give me more hints on how to start more than one Run/Debug Job to reference differient clients? I have not found the way to start both in the IDE.
We did start more than one job, go Run->Run/Debug, Create your first Job to be Run on Target OS-0 (You need to create different targets for the clients).
Start this job, at least to initialize, then pause it. Go again Run->Run/Debug, create 2nd job on Target OS-1.
Start it and pause it (or leave it runing) and resume the other job.
For later production use, i think the part with the watchproccess is pretty nice. There should be some API for that too, but not sure.