Commissioning a T2 Site

Introduction

This page is to be followed when injecting workflows to commission a Tier 2 Site

Subscribe block

Assign workflows via scripts

  • This will inject a workflow and assign it to agent cmssrv94 with team testbed-t1 and running on cmsweb testbed.
  • IMPORTANT:
    1. Since this is a test WF, you need to manually set the site in Normal operation mode (make sure it is not in drain) in the agent cmssrv94 which is for testing and separate from production. Otherwise, the WF will be stucked in acquired.
    2. The agent has a default threshold of 50 jobs/site. You need to change the threshold to the maximum slots the site allows to test the site at full capacity and evaluate the pledged resources.
    • ssh cmsdataops@cmssrv94
      source /data/admin/wmagent/env.sh
      ./config/wmagent/manage execute-agent wmagent-resource-control --normal --site-name=T1_AAA
      $manage execute-agent wmagent-resource-control --site-name=T1_AAA --pending-slots=xxx --running-slots=xxx --apply-to-all-tasks
      

  1. Login to any lxplus machine
    • ssh lxplus5.cern.ch
  2. Copy the necessary scripts from GITHub (it will copy the files to your global storage area for cern machines)
  3. Copy json file over and make modifications (the same things you would have configured in the web page)
  4. Set up your python to version 2.6
    • source /afs/cern.ch/cms/slc5_amd64_gcc461/external/python/2.6.4/etc/profile.d/init.sh
  5. Set up your proxy
    • source /afs/cern.ch/cms/LCG/LCG-2/UI/cms_ui_env.sh
    • voms-proxy-init -voms cms
  6. Create the WF request:
    • Replace all "T2_A" in the command with the name of the site you are commissioning
    • Change the Date "DDMMMYYYY" in the ProcessingString i.e.: 20SEP2013
    • python2.6 WmAgentScripts/ComissionT2Site/reqmgr.py --reqMgrUrl=https://cmsweb-testbed.cern.ch --json='{"createRequest":{"RequestString":"Commissioning_T2_A", "Campaign":"Commissioning_T2_A"},"assignRequest":{"SiteWhitelist":"T2_A", "AcquisitionEra":"Commissioning_T2_A", "ProcessingString":"DDMMMYYYY"}}' --configFile=MonteCarloFromGEN.json --createRequest --assignRequests

Check the workflow

  • The assigned WF is a long WF (>8 000 jobs) intended to take long time to complete so we don't have to keep assigning workflows.
  1. Check that the workflow does not have errors (very few are acceptable)
  2. Check Pending/Running jobs and Efficiency Plots:
  3. Check that all links to/from T1s are active at the site.
  4. Check that the site passes the Fallback tests
  5. When errors are found: check why and report them to the site or to the corresponding team.

If everything is successful

  • After few days testing you can abort the testing workflow, doesn't need to complete all, if jobs are successful.
  • Take site out of drain or skip INSTRUCTIONS
  • Announce that the site is ready for production in cms-comp-ops (Computing Operations Office Internal Discussions) <cms-comp-ops@cern.ch>
  • Remove the block used for testing

Remove block

  • To make a deletion request from PhEDEx, you should go the PhEdEx create deletion request page:
  • Put the block in the Data Items field
    • /QCD_HT-1000ToInf_TuneZ2star_8TeV-madgraph-pythia6/Summer12-START50_V13-v1/GEN#fafbeefe-5ef2-4b41-8ed5-1c21e578e3e1
  • Check your site in the Destination field.
  • Leave the Remove Subscriptions as default yes
  • Write your comments in the comment field (optional).
Edit | Attach | Watch | Print version | History: r17 < r16 < r15 < r14 < r13 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r17 - 2014-01-24 - JohnArtieda
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    CMSPublic All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback