TRIUMF Grid System Management Documentation

Notes

This is a running list of quick reminders about topics to document, to investigate, etc.

  • various LCG packages have no log rotation configuration.  This needs to be implemented.  A log rotation policy is needed.  The kickstart-ed configuration changes log rotation to a keep time of 3 months from 1 month.  However, logs need also to be permanently archived.  May 1st: done.
  • The compute elements and the lcggiis01 need larger /var/ partitions at install time - resolved for SL3 by moving /var files to /home/var and symlinking on the affected nodes.
  • yaim is not installed on the cluster; instead the files are mirrored at install time into /opt/lcg/yaim/   As of March 10th I have sync-ed the 2.3.1 version of yaim to the kickstart tree and to all nodes, preserving our own changes where necessary.  Timestamps of files in the functions subdir may not exactly correspond, but the file content is correct.
  • On March 10th I backed up all volatile and/or configuration files into lcfg:/export/data/admin/work/, and then rsync-ed those files to shrugged:/export/local2/grid-archive/
  • Need to chkconfig out rfiod on nodes