Filesystem problems on NIH Biowulf (Biowulf)

Biowulf users have probably noticed pauses while trying to
access their data directories over the past week.  In some
cases, jobs have failed with "Stale NFS file handle" errors.
We apologize for this disruption to your work.

During the downtime on April 26, a new version of IBM's
GPFS file system was installed on one of the Biowulf storage
systems, with the intent of improving reliability and
availability.  Unfortunately, beginning on May 29 something
in our job mix has uncovered a bug in the code which is
causing kernel panics in the fileservers.  Most pauses
encountered by users occur during fileserver failovers.

The Biowulf staff is actively pursuing a resolution to the 
problem with the storage vendor and with IBM.  We will
follow up this email with another once the problem has
been resolved.