Computing web : KCMS Computing projectIntroductionThis is the main page of the KCMS computing project. KCMS Computing workgroup activities:
![]() Computing managers are welcomed to present their status and give ideas to improve KCMS computing environment. Users in KCMS are also welcomed to share experiences in the analysis. Discussion items can be found in KCMS computing Task Force meeting ![]() ![]() General procedure to use KCMS computing resourcesPrerequirementsYou have to be registered as a KCMS member. To use CMS grid resources, you must register as a CMS member and create your grid certificate. Full registration procedure and getting grid certificate can be easily done when you visit CERN, but basically you can follow the procedures remotely in Korea. (It is best to visit CERN for your registration. Remote registration is slow and problematic in many cases.) Recommendations We recommend you to register to the official tier sites in Korea. There are three official tier sites, you can choose one depending on your location and group policy. Also it is free to register to all sites. Having account in tier site allows you to use local analysis farms, do easy-access of data files, store your large crab output files and share with your collaborators. You can also pull the dataset you are interested in to the site by PhEDEx? request (needs approval by site admin). Send registration form to site administrators with your information. Some site admins may ask you to be verified by your group leader. Check the followings for more information.
Using the KCMS computing resourcesAfter registration, you can setup your workarea in your directory. Usually, you are allowed to run test jobs only in the login machine. For jobs with heavy CPU usage, you have to use batch systems. Also user data files must be stored in dedicated storage machine (SE) which are accessable with srm or xrood protocols. For a standard cmsRun job submission, each sites provide instructions. You can submit your jobs with standard CRAB for grid jobs, local CRAB (if site is configured), create-batch script (ask Junghwan Goh) or submit jobs yourself.<----> Dataset for each site. |