How can I use a Linux computer as a Package Server without installing the Package Server Agent?
There is an alternative that may be implemented instead of using the Linux/Unix computer as a Package Server. You may want to keep this in mind. It is a workaround that has been implemented by several customers already.
The scenario is that the Unix/Linux computer is not actually a Package Server and the PS agent is not installed on it. The Unix/Linux computer is instead used as a "file store and access point," so to speak. This computer must be running web server services, such as Apache since unix/linux clients can only download packages via http from a valid URL.
To implement this:
Basically, we are telling the clients to get the package from a URL somewhere other than the NS server or a PS server. This can be any valid URL. Clients do not know the difference. They are simply being told to download the package from some URL. They will download both the snapshot and the codebase files from the NS server.
Since there is not a PS agent, the Unix/Linux box cannot automatically update the packages as they change on the NS server. There are a couple of options that can be implemented regarding package updating:
Note that the Unix/Linux computer, whether a PS or a simple file store, must be resolvable by hostname—either via DNS or by adding the Unix/Linux computer's IP address and hostname to the client's /etc/hosts file.