Nikhef Stoomboot
Nikhef Stoomboot tips and tricks
Copying from /eos
Use the following to copy from /eos in interactive session (good for small data):
setupATLAS
lsetup xrootd
lsetup xcache
voms-proxy-info -all
xrdcp --streams=4 --parallel=8 -r root://eosatlas.cern.ch//eos/atlas/atlaslocalgroupdisk/higgs/HHbbtautau/Run2Run3/EasyJetNTuples/Prod_v7_merged/HadHad /dcache/atlas/sjankovy/HHbbtautau/Prod_v7_merged
Setup for submitting jobs to copy from /eos
- Prerequisites. Run these in terminal:
setupATLAS
lsetup rucio
mkdir ~/.globus/proxies
voms-proxy-init -voms atlas -valid 192:00 -out ~/.globus/proxies/x509proxy
Be aware that the proxy is valid only for limited time (192 hours in this case), but the jobs can’t run longer than that anyway. You might need to resubmit the jobs after the jobs reach maximum runtime.
- Create
copy_job.shwith the following content./eos/atlas/atlaslocalgroupdisk/xxxis the source path and/dcache/atlas/${USER}/xxxis the destination path - make sure to put you big data to/dcache. If you can’t access the folder, write tostbc-admin@nikhef.nlto get access.
#!/bin/bash
export ATLAS_LOCAL_ROOT_BASE=/cvmfs/atlas.cern.ch/repo/ATLASLocalRootBase
source ${ATLAS_LOCAL_ROOT_BASE}/user/atlasLocalSetup.sh
# run this before submitting the job: `voms-proxy-init -voms atlas -valid 192:00 -out ~/.globus/proxies/x509proxy
export X509_USER_PROXY="~/.globus/proxies/x509proxy"
lsetup xrootd
lsetup xcache
voms-proxy-info -all
xrdcp --streams=4 --parallel=8 -r root://eosatlas.cern.ch//eos/atlas/atlaslocalgroupdisk/xxx /dcache/atlas/${USER}/xxx
- Create
copy_sub.shwith the following content:
#!/bin/bash
# HTCondor Submit File for xrdcp Transfer
# Executable script to run
executable = copy_job.sh
# Output, error, and log files
output = out/copy_$(Cluster).$(Process).log
error = err/copy_$(Cluster).$(Process).log
log = log/copy_$(Cluster).log
# Enable streaming for real-time output (requires HTCondor 8.9+)
stream_output = True
stream_error = True
# Request resources
request_cpus = 8
request_memory = 16GB
# Requirements and preferences
+JobCategory = "long"
+UseOS = "el9"
# Number of jobs to submit
queue 1
- Create directories for logs:
mkdir out err log
- Submit the job:
condor_submit copy_sub.sh