The purpose of this section is to describe how to enable a cluster of machines to run PROOF. Enabling PROOF means to configure and start a dedicated daemon on each machine running as master / worker, hereafter referred to as the servers.
The dedicated daemon - which is implemented as a plug-in for the multi-purpose xrootd daemon - processes PROOF-related requests on server nodes, which may come from the client or from a master on behalf of a client. The daemon is running on the PROOF server machines accepting connections on port 1093 (assigned by IAAA). It performs two tasks:
- Authenticates the requests: it checks that the request makes sense and comes from an authorized entity; the strength of the checks depends on the configuration settings;
- Starts a ROOT session and puts the client in connection with it; technically this is obtained by forking a child and execv'ing a proofserv application within it; proofserv instantiates a TProofServ which inherits from TApplication and runs the ROOT event loop.
There is still the possibility to run PROOF using the standalone proofd daemon; however, the development of proofd has been frozen in favour of the xrootd-based solution which provides additional coordination functionality and also the possibility to serve PROOF sessions and files with the same daemon.