This package consists of two programs, one called 'hpss2dpss', which is the program that transfers files from the HPSS to the DPSS, and the second program, which is called 'dpss2hpss', and conquently transfers the data from the DPSS to the HPSS. The installation of these two programs is fairly stright forward. For these programs to work there needs to be two 'nodes' installed in inetd. The modifications that have to be made are as follows:

add these lines to the '/etc/services' file:

hpss_node     15000/tcp

dp2hp_node    15001/tcp

and then to the file /etc/inetd.conf add the following lines:

hp2dp_node stream tcp nowait iss  /usr/local/DPSS/bin/hp2dp_node hp2dp_node

dp2hp_node stream tcp nowait iss  /usr/local/DPSS/bin/dp2hp_node dp2hp_node /usr/local/DPSS hpgw.lbl.gov

There are optional arguments to the two 'node' programs. The hp2dp_node program can be supplied with the location of your local DPSS directory (it defaults to /usr/local/DPSS). The dp2hp_node program can be supplied the location of you DPSS directory and the name of the machine runinng the local DPSS master program. The reason for giving the location of the local DPSS directory is that the program can then use the 'cache' of fastest interfaces to the DPSS and not have to rerun the 'find_best_path' algorithm again. (which can take several minutes, depending on network conditions)

You should always put in the cache directory (or make it /tmp) and the name of the machine running the DPSS master process

In the above example, it is assumed that there is a user named 'iss', if you don't have one, and don't mind the security risk, you can run the nodes as root a process. If you are security minded, you might want to create a user name iss or some such and run the process under that uid.