Fabric Management with Diskless Servers and Quattor on LHCb

At CERN LHCb experiment (Large Hadron Collider beauty), in order to reconstitute and filter events, a huge computing facility is required (currently ~1500 nodes). This computing farm, running SLC (Scientific Linux CERN, a Red Hat Enterprise Linux (RHEL) derivate is net booted and managed using Quattor to allow easy maintenance. LHCb is also using around 500 diskless Credit Card PCs (CCPCs) embedded in the front end electronic cards. To go farther than the limited Red Hat provided (and now deprecated) netboot tools, a new solution has been designed and will be shortly presented in this paper.