I agree; run the DB on the biggest (4GB RAM) box and maybe do master-slave to the next one (1GB RAM).
900MB is not a big DB for them to handle.
The cluster issue (I think) was due to each storage node (at least) must have the same amt of dedicated RAM, so you would have been limited to 512MB if using all 4.
I might be wrong about that, but I believe it was true in the early versions when the entire thing had to be in RAM.
The later version is disk backed, but its probably still true.
Check over at the mysql home forums
https://forums.mysql.com/ and/or main docs
http://dev.mysql.com/doc/refman/5.1/en/
If you can get any money at all or go through old PCs being thrown out etc, try to add more RAM to the other boxes.