I just moved to an all-Red Hat shop (HOORAY!!!) and I'm being faced with a situation that I'm not sure how to handle. We'd like to use the standard scheme under which common executables are stored on a central server and exported via NFS to all the clients, but I don't know how to make that work with RPM files. I'm used to using /opt or /usr/local as the net file system, installing software packages under their own directories with their own etc/ bin/ and lib/ subdirectories, and smoothing it all out with symlinks (so that the users don't have to have ridiculously long paths, and upgrades are as easy as installing the new package and moving the pointer). But RPM packages really want to install to the local system, with binaries in /usr/bin, config files in /etc, and so forth. I've tried using the --prefix and --root directives to rpm -i, but for at least one test case that doesn't work; either I get an error saying "this package is not relocatable", or (with some manual fiddling) I get the binaries and libs to go where they should but the software dies, unable to find its object libs. Maybe it's also worth mentioning that the NFS client systems are full-fledged systems, they don't net boot, and they have complete local setups (with /usr, /usr/local, and /var all housed on local disk). So at this point I figure I'm lacking some critical knowledge that would make this whole scheme workable. I assume that somebody is using RPM in an NFS environment, what might I be doing wrong? Thanks, -m -- Michael Jinks, IB Systems Administrator, Saecos Corporation _______________________________________________ Redhat-list mailing list [EMAIL PROTECTED] https://listman.redhat.com/mailman/listinfo/redhat-list
