Grid Computing Operations For The Cms Experiment At The Grif Tier-2

CMS moved into LHC data taking mode on November 2009. In order to store, distribute, process and analyze the experiment data, the CMS collaboration relies on a Grid-based system of distributed computing resources. Such resources are organized in "Tiers" and, in particular, Tier-2' sareregional level centersdevoted to support Montecarlo production and Physics analysis. GRIF, a distributed Tier-2 built of 6 computing centers in the Paris region, is one of such CMS-supporting sites. The main focus of GRIF is to support the important Physics analysis activity brought by a very active local CMS community.In this paper we describe the setup for CMS operation at GRIF, the lessons learned during the long commissioning period and the solutions adopted. We also present an over view of the CMS activities at the site both during the commissioning and the first data taking periods, pointing out successes and weaknesses.