Software Releases Management in the Trigger and Data Acquisition of Atlas Experiment - Integration, Building, Deployment, Patching

ATLAS is a general-purpose experiment in high-energy physics at Large Hadron Collider at CERN. ATLAS Trigger and Data Acquisition (TDAQ) system is a distributed computing system which is responsible for transferring and filtering the physics data from the experiment to mass-storage. TDAQ software is developed since 1998 by a team of few dozens developers. It is used for integration of all ATLAS subsystem participating in data-taking, providing framework and API for building the s/w pieces of TDAQ system. It is currently composed of more then 200 s/w packages which are available for ATLAS users in form of regular software releases. The s/w is available for development on a shared filesystem, on test beds and it is deployed to the ATLAS pit where it is used for data-taking. The paper describes the working model, the policies and the tools which are used by s/w developers and s/w librarians in order to develop, release, deploy and maintain the TDAQ s/w for the long period of development, commissioning and running the TDAQ system. In particular, the patching and distribution model based on RPM packaging is discussed, which is important for the s/w which is maintained for a long period on the running production system.