- aktuell - lists.lrz.de

Upd:Linux-Cluster: Completion of ICE maintenance delayed
by LRZ aktuell Sept. 12, 2011

Sept. 12, 2011

Changes to this message: Update Further update: Unfortunately we're still having problems with integrating the SLURM batch system with the ICE; before this is solved we cannot start user operation. We'll keep you up-to-date via this document when this is the case. Update: Due to the large workload involved in integrating the new resources into the Linux cluster, as well as the personnel and vacation situation at LRZ, returning the sgi ICE system to user operation is delayed. We hope to start production again on September 12 (this target date has also been updated below). Please note that the new MPP cluster has been opened for general user operation, and provides resources on a larger scale than the ICE. Submission of batch jobs is documented in http://www.lrz.de/services/compute/linux-cluster/batch_parallel/ ----------------------------------------------------------------------- The SGI ICE system will be unavailable for regular user operation between 22nd August and 11th of September, 2011. This is due to an operating system and batch queueing upgrade. After this extended maintenance, the system will be run in a SLURM cell of the Linux Cluster; please consult the cluster documentation at http://www.lrz.de/ services/compute/linux-cluster/ for upcoming details on this new batch software deployed on the parallel segments of the LRZ cluster. Submission of SGE jobs to the ICE will not work anymore after the upgrade. After the upgrade, it is recommended that you perform the following measures: * recompile and relink your programs using the new sgi MPT release * modify your batch scripts to use SLURM instead of SGE This information is also available on our web server http://www.lrz-muenchen.de/services/compute/aktuell/ali4083/ Reinhold Bader

1 0

Upd:Linux-Cluster: Completion of ICE maintenance delayed
by LRZ aktuell Sept. 12, 2011

Sept. 12, 2011

Changes to this message: Update further update: Unfortunately we're still having problems with integrating the SLURM batch system with the ICE; before this is solved we cannot start user operation. We'll keep you up-to-date via this document when this is the case. Update: Due to the large workload involved in integrating the new resources into the Linux cluster, as well as the personnel and vacation situation at LRZ, returning the sgi ICE system to user operation is delayed. We hope to start production again on September 12 (this target date has also been updated below). Please note that the new MPP cluster has been opened for general user operation, and provides resources on a larger scale than the ICE. Submission of batch jobs is documented in http://www.lrz.de/services/ compute/linux-cluster/batch_parallel/ ----------------------------------------------------------------------- The SGI ICE system will be unavailable for regular user operation between 22nd August and 11th of September, 2011. This is due to an operating system and batch queueing upgrade. After this extended maintenance, the system will be run in a SLURM cell of the Linux Cluster; please consult the cluster documentation at http://www.lrz.de/ services/compute/linux-cluster/ for upcoming details on this new batch software deployed on the parallel segments of the LRZ cluster. Submission of SGE jobs to the ICE will not work anymore after the upgrade. After the upgrade, it is recommended that you perform the following measures: * recompile and relink your programs using the new sgi MPT release * modify your batch scripts to use SLURM instead of SGE This information is also available on our web server http://www.lrz-muenchen.de/services/compute/aktuell/ali4083/ Reinhold Bader

1 0

HLRB-II down
by LRZ aktuell Sept. 12, 2011

Sept. 12, 2011

Dear users of HLRB-II, the HLRB-II is currently down. SGI is already working on the problem. Please use the alternate login node login2.hlrb2.lrz-muenchen.de to access your data and to submit jobs. We apologize for the inconvenience caused by this interrupt. This information is also available on our web server http://www.lrz-muenchen.de/services/compute/hlrb/aktuell/ali4105/ Hans-Georg Kleinhenz

1 0

Netzwartung am 13.9.
by LRZ aktuell Sept. 12, 2011

Sept. 12, 2011

1. TUM Gabelsberger Strasse 45 3.OG Zwischen 8:00 Uhr und 8:30 Uhr kommt es durch die Umschaltung des Anschlusses des Switches swr1-3bt auf 1 Gb/s zu einer Unterbrechung. 2. LMU Biologie II, Martinsried Bereits heute und auch morgen ist ganztaegig mit Unterbrechungen zu rechnen, da die Switches des Typs HP Procurve 4100 ausgetauscht werden. Diese Information finden Sie im WWW unter http://www.lrz-muenchen.de/services/netz/aktuell/ali4104/ Wolfgang Beyer

1 0

Upd:Linux-Cluster: Completion of ICE maintenance delayed
by LRZ aktuell Sept. 7, 2011

Sept. 7, 2011

Changes to this message: Update Update: Due to the large workload involved in integrating the new resources into the Linux cluster, as well as the personnel and vacation situation at LRZ, returning the sgi ICE system to user operation is delayed. We hope to start production again on September 12 (this target date has also been updated below). Please note that the new MPP cluster has been opened for general user operation, and provides resources on a larger scale than the ICE. Submission of batch jobs is documented in http://www.lrz.de/services/compute/linux-cluster/batch_parallel/ ----------------------------------------------------------------------- The SGI ICE system will be unavailable for regular user operation between 22nd August and 11th of September, 2011. This is due to an operating system and batch queueing upgrade. After this extended maintenance, the system will be run in a SLURM cell of the Linux Cluster; please consult the cluster documentation at http://www.lrz.de/ services/compute/linux-cluster/ for upcoming details on this new batch software deployed on the parallel segments of the LRZ cluster. Submission of SGE jobs to the ICE will not work anymore after the upgrade. After the upgrade, it is recommended that you perform the following measures: * recompile and relink your programs using the new sgi MPT release * modify your batch scripts to use SLURM instead of SGE This information is also available on our web server http://www.lrz-muenchen.de/services/compute/aktuell/ali4083/ Reinhold Bader

1 0

Upd:Linux-Cluster: Completion of ICE maintenance delayed
by LRZ aktuell Sept. 7, 2011

Sept. 7, 2011

Changes to this message: further delay Update: Due to the large workload involved in integrating the new resources into the Linux cluster, as well as the personnel and vacation situation at LRZ, returning the sgi ICE system to user operation is delayed. We hope to start production again on September 12 (this target date has also been updated below). Please note that the new MPP cluster has been opened for general user operation, and provides resources on a larger scale than the ICE. Submission of batch jobs is documented in http://www.lrz.de/services/compute/linux-cluster/batch_parallel/ ----------------------------------------------------------------------- The SGI ICE system will be unavailable for regular user operation between 22nd August and 11th of September, 2011. This is due to an operating system and batch queueing upgrade. After this extended maintenance, the system will be run in a SLURM cell of the Linux Cluster; please consult the cluster documentation at http://www.lrz.de/ services/compute/linux-cluster/ for upcoming details on this new batch software deployed on the parallel segments of the LRZ cluster. Submission of SGE jobs to the ICE will not work anymore after the upgrade. After the upgrade, it is recommended that you perform the following measures: * recompile and relink your programs using the new sgi MPT release * modify your batch scripts to use SLURM instead of SGE This information is also available on our web server http://www.lrz-muenchen.de/services/compute/aktuell/ali4083/ Reinhold Bader

1 0

Upd:Update: HLRB-II login partition back in production
by LRZ aktuell Sept. 6, 2011

Sept. 6, 2011

Changes to this message: Update Dear users of HLRB-II, the login partition of HLRB-II has been rebooted and is back in production. We apologize for the inconvenience caused by this interrupt. This information is also available on our web server http://www.lrz-muenchen.de/services/compute/hlrb/aktuell/ali4103/ Hans-Georg Kleinhenz

1 0

Upd:Linux-Cluster: Completion of ICE maintenance delayed
by LRZ aktuell Sept. 6, 2011

Sept. 6, 2011

Changes to this message: Further delay Update: Due to the large workload involved in integrating the new resources into the Linux cluster, as well as the personnel vacation situation at LRZ, returning the sgi ICE system to user operation is delayed. We hope to start production again on September 7 (this target date has also been updated below), probably in the afternoon. Please note that the new MPP cluster has been opened for general user operation, and provides resources on a larger scale than the ICE. Submission of batch jobs is documented in http://www.lrz.de/services/ compute/linux-cluster/batch_parallel/ ----------------------------------------------------------------------- The SGI ICE system will be unavailable for regular user operation between 22nd August and 7th of September, 2011. This is due to an operating system and batch queueing upgrade. After this extended maintenance, the system will be run in a SLURM cell of the Linux Cluster; please consult the cluster documentation at http://www.lrz.de/ services/compute/linux-cluster/ for upcoming details on this new batch software deployed on the parallel segments of the LRZ cluster. Submission of SGE jobs to the ICE will not work anymore after the upgrade. After the upgrade, it is recommended that you perform the following measures: * recompile and relink your programs using the new sgi MPT release * modify your batch scripts to use SLURM instead of SGE This information is also available on our web server http://www.lrz-muenchen.de/services/compute/aktuell/ali4083/ Reinhold Bader

1 0

HLRB-II login partition down
by LRZ aktuell Sept. 6, 2011

Sept. 6, 2011

Dear users of HLRB-II, the login partition of HLRB-II is currently down. To access your data and to submit jobs please use the alternate login partition login2.hlrb2.lrz-muenchen.de We apologize for the inconvenience caused by this interrupt. This information is also available on our web server http://www.lrz-muenchen.de/services/compute/hlrb/aktuell/ali4103/ Hans-Georg Kleinhenz

1 0

Supercomputer users: Moving from HLRB-II to the SuperMUC migration system
by LRZ aktuell Sept. 5, 2011

Sept. 5, 2011

Dear users of the HLRB-II at LRZ, The initial installation of the next-generation system, SuperMUC, is nearing regular user operation. As described in ALI 4069, the system has a computational power which is roughly equivalent to that of HLRB-II. LRZ has fixed the following time schedule for migration of all users from HLRB-II to the new system: 1. On the afternoon of September 13, all HLRB-II users gain access to the new machine 2. Between Sept. 13 and end of October, both HLRB-II and the new system will be online and available for batch processing 3. At the end of October, HLRB-II will be retired from user operation. Please note that at this time there will be an interruption of operation for all HPC systems for a few days because the data on the HOME file systems need to be migrated to a new and much more powerful file server; precise dates will be announced separately. Please note that beginning on September 10, LRZ will only receive limited maintenance for the HLRB-II. In particular, it may happen that the system loses partitions, or that the $OPT_TMP or $PROJECT file system is completely removed in case a critical part fails, if the replacement costs are considered unacceptably high. For this reason, we urgently recommend that you archive all still needed HLRB-II data which reside on the $OPT_TMP or $PROJECT file systems to TSM tape storage as soon as possible. A procedure for migrating your scratch and project data from HLRB-II to the new system is described in the backup and archiving document for the HPC systems; please look at the scenarios 2 and 3 for details on how to perform the data migration. The $HOME file system will access the same NAS space on both systems, hence you do not need to migrate any data from there yourself. This information is also available on our web server http://www.lrz-muenchen.de/services/compute/hlrb/aktuell/ali4102/ Reinhold Bader

1 0