Hardware failures increasing on /glade/p_old disk storage and Geyser/Caldera nodes

November 9, 2018

The frequency and severity of hardware failures on the Geyser and Caldera clusters and the /glade/p_old file system have increased as they approach end of life. These failures pose a risk to users who have not yet moved files that they need from /glade/p_old to one of the new storage spaces or have not begun to migrate their workflows to the new Casper data analysis and visualization cluster.

Several Geyser and Caldera nodes and /glade/p_old disks have already suffered irrecoverable hardware failures. They have been removed and will not be replaced. CISL storage engineers have been able to recover all files from failed /glade/p_old disks and will continue their data recovery efforts until December 15.  After that date, all files on /glade/p_old/ disks that fail will be lost permanently.

As announced previously, the /glade/p_old/ project and work spaces will be decommissioned December 31 and all files will be permanently removed from the system. CISL provides example scripts to help users move their files from old to new GLADE file spaces and documentation for using the new Campaign Storage archive. A tutorial planned for November 29 will demonstrate how to use the Globus CLI to work with Campaign Storage. Watch the Daily Bulletin for more information about the tutorial.