Project

General

Profile

GRID Resources

  • To set up ups products so they are visible locally or on grid worker nodes:
    . /grid/fermiapp/products/(experiment)/etc/setups.sh
  • Grid architecture for Intensity Frontier Submission
      General Configuration
          o if01 and if05 are condor submit nodes to fermigrids GPGrid OSG site.

          o Both if01 and gpsn01 run GlideinWMS as 3 users running 3 components
                + condor submitter running as root
                + the glidein factory running as unix user gfactory
                + the glidein frontend running as unix user gfrontend 

          o grid proxy certificates are refreshed every 24 hours by kcron running from user 
            dboxs account on the same machine that GlideinWMS is installed on 

      Starting/Stopping

          o Starting order is important. Suggested order to turn on:
               condor, gfactory, gfrontend.

          o condor starts automatically when machine is started. If you need to stop it for some reason:

            $ . /opt/condor/condor.sh
            $ . condor_off

            to turn back on:

            $ condor_on

          o to turn off/on the glidein factory

            $ ssh gfactory@if01
            $ ./stop_factory.sh
            $ ./restart_factory.sh

          o to turn off/on the glidein frontend

            $ ssh gfrontend@if01
            $ ./stop_fronted.sh
            $ ./restart_frontend.sh
  • The cpn and mvn commands require lock files at default location /grid/data/(experiment_gid)/LOCK
    commands for creating these:
    
    % mkdir -p /grid/data/(gid)
    % mkdir -p /grid/data/(gid)/LOCK
    % mkdir -p /grid/data/(gid)/LOCK/LOCKS
    % mkdir -p /grid/data/(gid)/LOCK/LOG
    % mkdir -p /grid/data/(gid)/LOCK/LOGS
    % mkdir -p /grid/data/(gid)/LOCK/QUEUE
    % mkdir -p /grid/data/(gid)/LOCK/STALE
    % chmod -R 775 /grid/data/(gid)/LOCK
    % find /grid/data/(gid)/LOCK -type d -exec chmod 775 {} \;
    % cd /grid/data/(gid)/LOCK
    %echo 50 > glimit #global open file limit
    %echo 5 > limit #open file limit
    %echo 3 > perf #required performance MB/sec before locking%echo '*'>PERF #measured performance MB/sec set by outside program like bluwatch
    %echo 10 > rate #target polling rate, per second
    %echo 10 > stale #ignore locks this old ( minutes )
    %echo 600 > staleq #ignore queue entries this old ( minutes )
    %echo 5 > wait #minimum retry delay ( seconds )