There are two types of proxies that the proxy manager deals with, managed and unmanaged proxies.
A managed proxy is a proxy that is started ``on demand'' by the proxy manager.
An unmanaged proxy, on the other hand, is started either at system boot time, or manually by a system administrator. The proxy manager is made aware of its existence, but no attempt is made by the proxy manager to start unmanaged proxies.
The command line options that can be specified to proxymngr are:
The proxy manager maintains a local configuration file describing the proxy services available. This configuration file is installed in /usr/X11R6/lib/X11/proxymngr/pmconfig during the installation of proxymngr. The location of the configuration file can be overwritten using the -config command line option.
Aside from lines starting with an exclamation point for comments, each line of the configuration file describes either an unmanaged or managed proxy service.
For unmanaged proxies, the format is:
service-name is the name of the unmanaged proxy service, and must not contain any spaces, for example ``XFWP''. service-name is case insenstive.
proxy-address is the network address of the unmanaged proxy. The format of the address is specific to the service-name. For example, for the ``XFWP'' service, the proxy-address might be ``firewall.x.org:100''.
If there is more than one entry in the config file with the same unmanaged service-name, the proxy manager will try to use the proxies in the order presented in the config file.
For managed proxies, the format is:
service-name is the name of the managed proxy service, and must not contain any spaces, for example ``LBX''. service-name is case insensitive.
command-to-start-proxy is the command executed by the proxy manager to start a new instance of the proxy. If command-to-start-proxy contains spaces, the complete command should be surrounded by single quotes. If desired, command-to-start-proxy can be used to start a proxy on a remote machine. The specifics of the remote execution method used to do this is not specified here.
Here is a sample configuration file:
! proxy manager config file ! ! Each line has the format: ! <serviceName> managed <startCommand> ! or ! <serviceName> unmanaged <proxyAddress> ! lbx managed /usr/X11R6/bin/lbxproxy ! ! substitute site-specific info xfwp unmanaged firewall:4444
When the proxy manager gets a request from xfindproxy (or another similar client), its course of action will depend on the service-name in question.
For a managed proxy service, the proxy manager will find out if any of the already running proxies for this service can handle a new request. If not, the proxy manager will attempt to start up a new instance of the proxy (using the command-to-start-proxy found in the config file). If that fails, an error will be returned to the caller.
For an unmanaged proxy service, the proxy manager will look in the config file to find all unmanaged proxies for this service. If there is more than one entry in the config file with the same unmanaged service-name, the proxy manager will try to use the proxies in the order presented in the config file. If none of the unmanged proxies can satisfy the request, the proxy manager will timeout for a configurable amount of time (specified by -timeout or default of 10) and reattempt to find an unmanaged proxy willing to satisfy the request. The number of retries can be specified by the -retries argument, or a default of 3 will be used. If the retries fail, the proxy manager has no choice but to return an error to the caller (since the proxy manager can not start unmanaged proxy services).
-timeout and -retries is not implemented in proxymngr.
proxymngr does not utilize the ``options'' and ``host'' fields in the proxy management protocol GetProxyAddr request.