> I don't think I completely understand the procedure you are describing,
Ah, yes, sorry I did miss out a critical bit of what I am doing! I am compiling on a stable/13 box, and then using NFS to mount /usr/src and /usr/obj onto the test machine. On that machine I am doing installkernel and installworld to upgrade it. > just a note that guessing on the /tmp/legacy/ in path it's a "native" > tool that runs on build system and was compiled against the libraries on > the build system (unlike real binaries built against the libraries for > the target system). Yes, thast the conclusion I came to. I am also not sure that it ever actually runs that binary - the failure is right at the start of installworld where it checks to make sure all the shared libraries are here. Possibly what I am doing is not supported, but I am a bit surprised as my usual upgrade procedure is to build on one machine and then either NSF mount, or rsync, the files to the boxes on which I wish to install them. -pete.
