|
|
|
|
Basic information about the HLRN system
Revision History: Revision 2.5, Published 2006/05/10 10:24:26 (UTC) by Wolfgang Baumann Table of Contents Abstract This document gives brief information about configuration and usage of the HLRN system. It serves as a starting point and gives a number of important Links. Further information is given in more detail in the collection of HLRN user documentation (HLRNdoc). The most important documents are contained in the HLRN Quickstart Guide. In addition a number of software packages and their uses are described - see the collection about Software, Libraries, etc. The following system is in operation:
Resources are primarily allocated to projects which have been successfully reviewed by the Scientific Board of HLRN (Wissenschaftlicher Ausschuss des HLRN). Details about the procedure and important hints are given on their web pages. These documents are written in German. Access to the HLRN system is given after successfully applying for an user account via the web-based form Beantragen von allgemeinen Benutzerkennungen und Betriebskennungen on Upon correct completion of the form, an email is sent which must be printed, signed by the applicant, and presented to your computing center (see Section 3, "Which one is "Your Computing Center"?"). Initially accounts have a limited period of validity and are allocated very limited resources unless they are integrated within a project (see above). If you have questions concerning the current status of your application or if you want to have your password changed by administrative means (e.g. in cases of suspected misuse or if simply forgotten): The user administration for the whole HLRN system can be contacted during normal operation hours
The HLRN Quickstart Guide, the document Allocations, Accounts, ..., and the collection Access and Environment contain further information.
Both complexes are managed by one LoadLeveler system. Jobs may direct the target complex to run themselves independently from the complex at which they are submitted (for valid job classes see Batch System Partitions and Classes in LoadLeveler - The IBM SP Batch Queuing System).
More detailed information is given in File Systems and Storage in The HLRN Quickstart Guide. Concerning user administration, LoadLeveler, file systems and software the login servers berni1 and berni2 as well as the compute nodes breg01 - breg16 (complex at ZIB) form one system; each compute node is configured as full SMP system with 32 processors and 64 GByte of memory (except breg11 with 128 GByte and breg12 with 256 GByte of memory). The file system /perm/ is accessible only on the data server bdata. The same details are true for the login servers hanni1, hanni2 and the compute nodes hreg01 - hreg16 (complex at RRZN). The intended use of bdata and hdata is interactive pre- and postprocessing of jobs as well as data transfer to and from external systems. Connecting call e.g.: ssh <userid>@bdata.hlrn.deThe current operational status of the complexes in Berlin and Hannover may be observed via the web. See the respective Status page. Important News are published on the HLRN System News website. 2003-2008 © Norddeutscher Verbund für Hoch- und Höchstleistungsrechnen (HLRN) |
|||||||||||||||||||||||||||||||||
hlrn.de>