clip-hpc/grid
Agent skills for grid computing at the Vienna BioCenter: grid certificates and VOMS proxies, CVMFS experiment software, EOS and XRootD transfers, grid job submission to the Vienna Tier-2 (Hephy-Vienna), and failure triage.
Read, write, and transfer data on EOS and grid storage from CBE.next: the /eos FUSE mount, the eos CLI (eos ls, eos quota), CERNBox paths under /eos/vbc, XRootD with xrdcp and xrdfs against root://eos.grid.vbc.ac.at, WebDAV on https://eos.grid.vbc.ac.at:8443, gfal-copy and gfal-ls, adler32 checksums, and staging EOS data for Slurm jobs. Use for EOS, XRootD, root:// URLs, davs://, grid storage elements, and moving data between EOS and /groups or /scratch. Use cbe-data for other CBE.next storage and transfers, grid-identity for certificates, proxies, and EOS mapping, and grid-triage for failing transfers.
Set up and check a grid identity for ALICE, CMS, or Belle II work at the Vienna BioCenter: request a CERN grid certificate, convert the .p12 into ~/.globus/usercert.pem and userkey.pem, fix file permissions, join a VO (VOMS or IAM), create and inspect VOMS proxies with voms-proxy-init and voms-proxy-info, use X509_USER_PROXY in Slurm jobs on CBE.next, renew proxies, and get the certificate DN mapped for EOS. Use for grid certificate, DN, VOMS, proxy, FQAN, and X509_* questions; use grid-data for transfers, grid-jobs for submission, and grid-triage for failures.
Submit and follow grid jobs for ALICE, CMS, and Belle II, in general or targeted at the Vienna Tier-2 (WLCG site Hephy-Vienna, CMS T2_AT_Vienna, ALICE Vienna, Belle II LCG.HEPHY.at, HTCondor-CE ce-1.grid.vbc.ac.at): ALICE JAliEn with alien.py and JDL, CMS CRAB with crab submit and Site.whitelist, Belle II gbasf2 and DIRAC, direct HTCondor-CE tests with condor_ce_trace and grid universe submit files, and what the CBE.next partition grid with QOS grid_generic is for. Use when choosing between grid submission and local Slurm jobs, or moving legacy grid_generic workflows. Use cbe-next for ordinary sbatch jobs, grid-identity for proxies, grid-software for experiment environments, and grid-triage for failed or stuck grid jobs.
Set up HEP and grid experiment software from CVMFS on CBE.next: ALICE O2Physics and AliPhysics with alienv from /cvmfs/alice.cern.ch, CMS CMSSW with cmsset_default.sh, cmsrel, cmsenv, and SCRAM_ARCH from /cvmfs/cms.cern.ch, Belle II basf2 with b2setup from /cvmfs/belle.cern.ch, ROOT and LCG views from /cvmfs/sft.cern.ch, older OS releases in containers from /cvmfs/unpacked.cern.ch, and grid client tools such as voms-proxy-init, xrdcp, gfal, and alien.py. Use when an experiment environment or setup script fails or a CVMFS repository looks missing. Use cbe-software for modules, EESSI, conda, and non-HEP software, and grid-identity for certificates and proxies.
Diagnose grid failures for ALICE, CMS, and Belle II users at the Vienna BioCenter: voms-proxy-init errors, expired proxies or VOMS attributes, wrong VO or FQAN, unmapped certificate DN, CA, CRL, or certificate verification errors, xrdcp, xrdfs, gfal, and WebDAV failures against eos.grid.vbc.ac.at such as Auth failed, permission denied, or checksum mismatch, hanging or stale /eos FUSE mounts, missing CVMFS repositories, and held, failed, or stuck CRAB, JAliEn, gbasf2, or HTCondor-CE jobs. Hands over to cbe-ticket when the cause is on the site's side. Use cbe-triage for ordinary Slurm job failures.