A data-driven approach to configure permanent tasks in LHCb Online farm

During the boot sequence, the LHCb Online system consisting of about 1700 CPUs needs to configure roughly 2 dozen different types of standard, off the shelf processor units, each running a dedicated set of permanent processes. In this report, a data driven approach is presented where these node types and the corresponding processes are modeled using a relational database. Starting from the database model, a set of utility applications were derived. These include a command line interface, a web based graphical user interface and a program to verify the existence of the processes on a given node. Future extensions can easily be implemented using the provided application programming interface.