AtticRWG

2007

• Need to fix http://mwt2.uchicago.edu/atlas-analysis/atlas-workbook (was this plone?)

10/19/06

• Doing ITB 0.5.1 install on OsgITBDev
• Crontab man page, http://docsrv.sco.com/cgi-bin/man/man?crontab+1
• Vi editor page from uiuc: http://www.uic.edu/depts/accc/software/unixgeneral/vi101.html

10/10/06

• http://twiki.mwt2.org/bin/view/Admins/FaceToFaceOct06
• Teraport front-page twiki needs work.
• MPI on Teraport-need a plan.

9/14/06

• Looking over ITB install on development cluster, at: http://twiki.mwt2.org/bin/view/Admins/OsgITBDev
• Plan to install ITB 0.5.1 there.

6/8/06

• DQ2 operations - see http://twiki.mwt2.org/bin/view/Admins/DonQuijote2DQ2
• Marty's task focus -
  • Update CRL: http://gridui01.usatlas.bnl.gov:25880/server/pandamon/query?dash=prod.
    • OSG 0.4.1 install instructions - http://gridui01.usatlas.bnl.gov:25880/server/pandamon/query?dash=prod. This is installed on tier2-02 at /opt/osg/. "osg" alias was incorrectly pointing to an old installation. Go to tier2-mgt, and edit .bash_profile file. Need to make this available to all the headnodes (need to redo softlinks).
      • Marty's script push /.bash_profile works. (Don't use Rocks on headnodes.)
    • A recent pacman is on /share/adm. (version 3.18.4)
    • pacman -update CA-Certificates. This seems to work.
• Nanohub and CDF VO's - how to enable?
  • Accounts already created. How were these names chosen?
  • How do I check the GUMS server?
  • gums-host generateGridMapfile this works from t2dev, but not on tier2-02.
  • confusion about old grid installations - /share/grid has a year old OSG install. I say lets delete it.
  • now, questions about what should be exported.
  • remove mount point for /share/grid in /etc/auto.share/ - this is done on tier2-mgt. (comment out for now)
  • push /etc/auto.share out to headnodes.
  • 411put will put it out to the workernodes.
  • reload the automounter's configuration clusterwide. services autofs reload.
  • onall -a does all the headnodes. onall is a script written by Marty, its in /share/adm.
  • Marty also uses /var/adm, for node specific configuration and scripts.
  • Then: =cluster-fork /etc/init.d/autofs reload =. automounter is reloading its configuration.
  • Now - osg workernode client is installed someplace, and is exported everywhere.
  • look at /opt/osg/gums for more config files.
• Twiki opening, can't find the preferences setting. - wait for Greg
• Condor behavior
  • update to 6.7.20 (current, by the website)
  • See notes at http://pgl.uchicago.edu/twiki/bin/view/Tier2/ConAdm
• ITB cluster diagnostics
  • Grid exerciser reports - http://www.cs.wisc.edu/condor/tools/exerciser/reading_report.html, http://www.cs.wisc.edu/condor/tools/exerciser/
  • Globus error 12
  • log in as a user, reproduce the error.
  • /etc/xinetd.d/globus-gatekeeper
  • reboot.. still fails.
  • looking at prima-authz.conf
• Two nodes require manual assembly of software RAIDS. tier2-04 and atlashome. MDADM. Can we generate a recipe to recover in case either of these crash. Marty will generate a recipe. .
• Marty will document what was done to put condor uprade into rocks on dev cluster.