Cluster Resources

Newsletter – May 2005

Optimizing your Compute Resources

NEWS/ANNOUNCEMENTS

1. DATA STAGING FOR MOAB COMING SOON

2. NEW PATCHES FOR TORQUE

3. MOAB ACCESS PORTAL FOR GRID NOW IN ALPHA STAGE

4. JOB TEMPLATES NOW SUPPORTED IN MOAB CLUSTER MANAGER

5. DID YOU KNOW: NODE SETS

*************************************************************************

DATA STAGING FOR MOAB COMING SOON

The next release of Moab Workload Manager^® (v4.2.4), scheduled for release in July 2005, will include intelligent handling of job data staging. This feature can increase compute, networking and overall cluster performance as much as 40 percent, depending on the workload and cluster infrastructure.

Several levels of data staging functionality will be supported depending on the amount of control and information available to Moab. The most basic level of data staging will schedule a job only after any input files are fully staged. The most advanced level of data staging, which will be available in the subsequent release, will fully schedule the data staging operation so that it completes just before the scheduled start time of the job. Cluster Resources is currently Beta testing the new functionality with several partners and expects to have a full GA release this year, possibly by July or later.

*************************************************************************

NEW PATCHES FOR TORQUE*

TORQUE Resource Manager* continues to receive new patches submitted by the user community. Some of the highlights of TORQUE's newest patches include:

TORQUE 1.2.0p4

-Extended “job prolog” to include jobname, resource, queue and account info (courtesy of University of Maine)
-Added support for Darwin 10.4 (courtesy of University of Maine)
-Fixed suspend/resume for MPI jobs
-Added support for epilog.precancel to enable local job cancellation handling
-Fixed build for case insensitve filesystems
-Fixed relative path based Makefiles for xpbsmom

TORQUE 1.2.0p3

-Enable multiple server to mom communication
-Fixed node reject message overwrite issue
-Enable pre-start node health check (Courtesy of BOEING)
-Fixed pid scanning for RHEL3 (Courtesy of VPAC)
-Added improved vmem/mem limit enforcement and reporting (Courtesy of UMU)
-Added submit filter return code processing to qsub

On behalf of the user community we would also like to thank all those not mentioned above who helped find and fix problems and test new patches. To learn more about TORQUE visit: www.clusterresources.com/products/torque/. To join TORQUE's user community go to: www.clusterresources.com/mailing.shtml.

*************************************************************************

MOAB ACCESS PORTAL FOR GRID NOW IN ALPHA STAGE

Moab Access Portal for Grids^® provides expanded capabilities for Moab Grid Suite^®. Now in an Alpha stage, this new version of Moab Access Portal allows end users to not only manage their workload on the cluster level, but also to run workload on grid resources. Running grid jobs is as easy as selecting whether a job should be placed in a grid queue or run on a particular cluster. Grid resources appear to end-users as virtual nodes and can be monitored just like local resources.

Moab Access Portal for Grids provides a seamless transition for cluster-centric users to move to the more powerful grid environments. To view the cluster version of Moab Access Portal, go to: www.clusterresources.com/eval and click on the Portal Online Demo link at the bottom of the page. Expect to see Moab Access Portal for Grids released to the public sometime during the third quarter of this year.

*************************************************************************

JOB TEMPLATES NOW SUPPORTED IN MOAB CLUSTER MANAGER^®

The latest Moab Cluster Manager^® version 2.2.1 now has support for job templates. This convenient new feature allows users to save their job submission settings for later use. Using job templates, a user can focus on the job itself and ignore the repetitious and often tedious task of entering the same job submission script multiple times.

Users can save personal job templates that are available exclusively to the creator. Moab Cluster Manager also supports global job templates. When administrators or users create global job templates, the job submission settings are saved to a global location where other users can access the templates. This allows them to carefully set up a job submission template that dozens of other users can simply load and apply. Users can modify any settings and then submit the job.

*************************************************************************

DID YOU KNOW: NODE SETS

In clusters, most parallel jobs run only as fast as the slowest node to which they have been allocated. As sites increase their cluster size and integrate newer compute nodes that are not homogeneous with earlier nodes, average job efficiency generally decreases. Using Moab Workload Manager^®, node sets may be programmed to increase job efficiency by specifying that a job run on homogeneous nodes. The node set feature improves job efficiency by allowing jobs to request sets of common resources (processor speed, memory, network interfaces or locally defined node attributes) without specifying exactly what resources are required. Furthermore, the node set feature can guide jobs to the nodes on which they will have the best performance.

For example, an I/O intensive job may run too slowly on a cluster's slowest nodes, but may waste processing efficiency on a cluster's fastest nodes. In this case, one may set a range of processor speeds the job would run most efficiently on, and the job will only be run on processors within the specified range of speed.

To learn more about node sets and how to apply them to Moab Workload Manager, visit www.clusterresources.com/products/mwm/docs/8.3nodesetoverview.shtml, or to learn how to apply node sets on Maui Cluster Scheduler, visit:www.clusterresources.com/products/maui/docs/8.3nodesetoverview.shtml.

Moab Cluster Suite^® , Moab Grid Suite^®, Moab Workload Manager^®, Moab Cluster Manager^®, and Moab Access Portal^® are trademarks or registered trademarks of Cluster Resources Incorporated_TM.

* This product includes software developed by NASA Ames Research Center, Lawrence Livermore National Laboratory, and Veridian Information Solutions, Inc. Visit www.OpenPBS.org for OpenPBS software support, products, and information. TORQUE is neither endorsed by nor affiliated with Altair Grid Solutions, Inc.


	Learn More About Us Product Information Access Portal Demo Highlighted Customers Training Documentation News Download Moab Cluster Suite ·Customer Download ·Eval Download ·Access Portal Demo Moab Grid Suite ·Customer Download ·Eval Download Maui TORQUE Purchase Products Support Other Services Authorized Resellers Support Customer Support Purchase Support Free Eval Support Documentation Mailing List Contact Us Product Information Sales Technical Support Free Eval Support Partner Press
Home Products Services Resources Partners Press Company	Newsletter – May 2005 Optimizing your Compute Resources NEWS/ANNOUNCEMENTS 1. DATA STAGING FOR MOAB COMING SOON 2. NEW PATCHES FOR TORQUE 3. MOAB ACCESS PORTAL FOR GRID NOW IN ALPHA STAGE 4. JOB TEMPLATES NOW SUPPORTED IN MOAB CLUSTER MANAGER 5. DID YOU KNOW: NODE SETS *********************************************************************** DATA STAGING FOR MOAB COMING SOON The next release of Moab Workload Manager^® (v4.2.4), scheduled for release in July 2005, will include intelligent handling of job data staging. This feature can increase compute, networking and overall cluster performance as much as 40 percent, depending on the workload and cluster infrastructure. Several levels of data staging functionality will be supported depending on the amount of control and information available to Moab. The most basic level of data staging will schedule a job only after any input files are fully staged. The most advanced level of data staging, which will be available in the subsequent release, will fully schedule the data staging operation so that it completes just before the scheduled start time of the job. Cluster Resources is currently Beta testing the new functionality with several partners and expects to have a full GA release this year, possibly by July or later. *********************************************************************** NEW PATCHES FOR TORQUE* TORQUE Resource Manager* continues to receive new patches submitted by the user community. Some of the highlights of TORQUE's newest patches include: TORQUE 1.2.0p4 -Extended “job prolog” to include jobname, resource, queue and account info (courtesy of University of Maine) -Added support for Darwin 10.4 (courtesy of University of Maine) -Fixed suspend/resume for MPI jobs -Added support for epilog.precancel to enable local job cancellation handling -Fixed build for case insensitve filesystems -Fixed relative path based Makefiles for xpbsmom TORQUE 1.2.0p3 -Enable multiple server to mom communication -Fixed node reject message overwrite issue -Enable pre-start node health check (Courtesy of BOEING) -Fixed pid scanning for RHEL3 (Courtesy of VPAC) -Added improved vmem/mem limit enforcement and reporting (Courtesy of UMU) -Added submit filter return code processing to qsub On behalf of the user community we would also like to thank all those not mentioned above who helped find and fix problems and test new patches. To learn more about TORQUE visit: www.clusterresources.com/products/torque/. To join TORQUE's user community go to: www.clusterresources.com/mailing.shtml. *********************************************************************** MOAB ACCESS PORTAL FOR GRID NOW IN ALPHA STAGE Moab Access Portal for Grids^® provides expanded capabilities for Moab Grid Suite^®. Now in an Alpha stage, this new version of Moab Access Portal allows end users to not only manage their workload on the cluster level, but also to run workload on grid resources. Running grid jobs is as easy as selecting whether a job should be placed in a grid queue or run on a particular cluster. Grid resources appear to end-users as virtual nodes and can be monitored just like local resources. Moab Access Portal for Grids provides a seamless transition for cluster-centric users to move to the more powerful grid environments. To view the cluster version of Moab Access Portal, go to: www.clusterresources.com/eval and click on the Portal Online Demo link at the bottom of the page. Expect to see Moab Access Portal for Grids released to the public sometime during the third quarter of this year. ********************************************************************* JOB TEMPLATES NOW SUPPORTED IN MOAB CLUSTER MANAGER^® The latest Moab Cluster Manager^® version 2.2.1 now has support for job templates. This convenient new feature allows users to save their job submission settings for later use. Using job templates, a user can focus on the job itself and ignore the repetitious and often tedious task of entering the same job submission script multiple times. Users can save personal job templates that are available exclusively to the creator. Moab Cluster Manager also supports global job templates. When administrators or users create global job templates, the job submission settings are saved to a global location where other users can access the templates. This allows them to carefully set up a job submission template that dozens of other users can simply load and apply. Users can modify any settings and then submit the job. ********************************************************************* DID YOU KNOW: NODE SETS** In clusters, most parallel jobs run only as fast as the slowest node to which they have been allocated. As sites increase their cluster size and integrate newer compute nodes that are not homogeneous with earlier nodes, average job efficiency generally decreases. Using Moab Workload Manager^®, node sets may be programmed to increase job efficiency by specifying that a job run on homogeneous nodes. The node set feature improves job efficiency by allowing jobs to request sets of common resources (processor speed, memory, network interfaces or locally defined node attributes) without specifying exactly what resources are required. Furthermore, the node set feature can guide jobs to the nodes on which they will have the best performance. For example, an I/O intensive job may run too slowly on a cluster's slowest nodes, but may waste processing efficiency on a cluster's fastest nodes. In this case, one may set a range of processor speeds the job would run most efficiently on, and the job will only be run on processors within the specified range of speed. To learn more about node sets and how to apply them to Moab Workload Manager, visit www.clusterresources.com/products/mwm/docs/8.3nodesetoverview.shtml, or to learn how to apply node sets on Maui Cluster Scheduler, visit:www.clusterresources.com/products/maui/docs/8.3nodesetoverview.shtml. Moab Cluster Suite^® , Moab Grid Suite^®, Moab Workload Manager^®, Moab Cluster Manager^®, and Moab Access Portal^® are trademarks or registered trademarks of Cluster Resources Incorporated_TM. * This product includes software developed by NASA Ames Research Center, Lawrence Livermore National Laboratory, and Veridian Information Solutions, Inc. Visit www.OpenPBS.org for OpenPBS software support, products, and information. TORQUE is neither endorsed by nor affiliated with Altair Grid Solutions, Inc.
© 2001-2007 Cluster Resources, Inc.