Oct. 3, 2012
10:52 a.m.
Changes to this message: Update 5th Update (Oct 3, 12:50) The cause of the GPFS failure on Saturday, Sep 29 has been identified as a software problem triggered by data migration to additionally installed disk space. After performing the data migration in offline mode, the system has now been returned to normal user operation again. Please produce a new login shell to properly set the SCRATCH and WORK environment variables in order to access the GPFS data from the login nodes. Submission of LoadLeveler jobs that use GPFS should now also work again. This information is also available on our web server http://www.lrz-muenchen.de/services/compute/supermuc/aktuell/ali4424/ Markus Mueller