I'm using some old hardware (and vmware instances) to verify some SuSE 9.0 autoinstalls. Some only have 128MB of RAM. While I wouldn't attempt to run any production services on these boxes, they're a good test of my final scripts and processes. I've run across a problem in the way the rpm's are installed using a network (http) source. By pushing a virtual console during the actual install, it appears that the rpm command is caching the rpm to be installed in / (the ramfs) vs. the mounted hardrive. So basically, this limits the size of an rpm that can be installed from a network source to roughly Physical Ram - ( kernel + cramfs installer ram footprint), which works out to roughly 52MB on a 128M system. With less than 128M there are other concerns, so that's my bare minimum for testing installs. Is there a switch I can pass AutoYaST2 somewhere to have it use the installing system's mounted /var/tmp vs. the ramfs? There are some pretty big RPM's out there, like OpenOffice (who's integration I need to test :^). Thanks, I realize this is a corner case most people would never hit on modern hardware (HTTP install + 128M RAM). Lee