There are a couple of ways. If you are talking system files and OS images, check out Warewulf, http://warewulf.lbl.gov/trac
I have found placing packages in /opt/ and sharing via NFS works well for small to medium systems. I make a directory called /opt/etc/ld.so.conf.d/ and place the package library conf files in this directory. I then add "include /opt/etc/ld.so.conf.d/*.conf" to the node /etc/ld.so.conf file. That way, I don't need to reboot or change anything on the nodes when I add a new package. It is installed in NFS and then I issue a "pdsh ldconfig" Actually I package everything in rpms, so all this gets done automatically when things are installed. I use Modules and install a Modules file for each package that sets paths etc., it can override the ld.so.conf files if needed by using a LD_LIBRARY_PATH. I also do some sshd_conf twiddling so user Module state is preserved across logins to cluster nodes. -- Doug > Hello all, > > I was wondering what solutions other use for easy software installation > across clusters. Is there any method that is generally accepted to be the > most effective for making sure that all the software is consistent across > each node? > > Thanks, > Trevor > _______________________________________________ > Beowulf mailing list, [email protected] sponsored by Penguin Computing > To change your subscription (digest mode or unsubscribe) visit > http://www.beowulf.org/mailman/listinfo/beowulf > > -- > Mailscanner: Clean > -- Doug -- Mailscanner: Clean _______________________________________________ Beowulf mailing list, [email protected] sponsored by Penguin Computing To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
