IBM Spectrum LSF 10.1 Fix 524550 Readme File

Abstract 

RFE#125752. Enhance LSF GPU MPS integration to enable the shared use GPU through MPS for the same user with the same job requirement.

 

Description

Readme documentation for IBM Spectrum LSF 10.1 Fix 524550 including installation-related instructions, prerequisites and co-requisites, and list of fixes.

 

LSF now allows you to run multiple NVIDIA Multi-Process Service (MPS) daemons on a host for GPU jobs and allows you to share these MPS daemons across multiple GPU jobs.

 

To use this enhancement, LSB_GPU_NEW_SYNTAX=extend must be defined in the lsf.conf file to enable MPS daemons.

 

To define the behavior of running and sharing multiple MPS daemons, this fix adds new values to the existing mps keyword in the GPU resource requirements string (that is, the bsub -gpu command option, LSB_GPU_REQ parameter in the lsf.conf file, and the GPU_REQ parameter in the lsb.queues and lsb.applications files):

 

-gpu “mps=default,shared”

LSF starts MPS daemon per host for multiple jobs submitted by the same user with the same job submission requirements.

-gpu “mps=per_socket,shared”

LSF start MPS daemon per socket for multiple jobs submitted by the same user with the same job submission requirements.

-gpu “mps=per_gpu,shared”

LSF start MPS daemon per GPU for multiple jobs submitted by the same user with the same job submission requirements.

 

For optimal performance, use mode=exclusive_process together with MPS shared keyword.

 

Readme file for: IBM® Spectrum LSF

Product/Component Release: 10.1

Update Name: Fix 524550 

Fix ID: LSF-10.1-build524550 

Publication date: 10 July 2019 

Last modified date:  10 July 2019 

Contents:

1.     List of Fixes

2.     Download location 

3.     Products or components affected

4.     System requirements 

5.     Installation and configuration

6.     List of files

7.     Product notification

8.     Copyright and trademark information

 

1.   List of Fixes

RFE#125752

 

2.   Download Location

Download Fix 524550 from the following location: http://www.ibm.com/eserver/support/fixes/

 

3.   Products or components affected

Affected components include: LSF/mbschd LSF/mbatchd LSF/sbatchd LSF/res LSF/bhist LSF/bjobs LSF/bmod LSF/bresize LSF/bsub

 

4.   System requirements

linux2.6-glibc2.3-x86_64

lnx3.10-lib2.17-x86_64

linux3.10-glibc2.17-ppc64le

 

5.   Installation and configuration

5.1    Before installation

         (LSF_TOP=Full path to the top-level installation directory of LSF.)

         1)    Log on to the LSF master host as root

         2)    Set your environment:

         -      For csh or tcsh: % source LSF_TOP/conf/cshrc.lsf

         -      For sh, ksh, or bash: $ . LSF_TOP/conf/profile.lsf

 

5.2    Installation steps

         1)    Go to the patch install directory: cd $LSF_ENVDIR/../10.1/install/

         2)    Copy the patch file to the install directory $LSF_ENVDIR/../10.1/install/

         3)    Run badmin hclose all

         4)    Run badmin qinact all

         5)    Run patchinstall: ./patchinstall <patch>

 

5.3    After installation

         1)    Log on to the LSF master host as root

         2)    Run lsadmin resrestart all

         3)    Run badmin hrestart all

         4)    Run badmin mbdrestart -s

         5)    Run badmin hopen all

         6)    Run badmin qact all

 

5.4    Uninstallation

        1)    Log on to the LSF master host as root

        2)    Run badmin hclose all

        3)    Run badmin qinact all

        4)    Go to the patch install directory: cd $LSF_ENVDIR/../10.1/install/

        5)    Run ./patchinstall -r <patch>

        6)    Run lsadmin resrestart all

        7)    Run badmin hrestart all

        8)    Run badmin mbdrestart -s

        9)    Run badmin hopen all

      10)    Run badmin qact all

 

6.   List of files 

mbschd

mbatchd

sbatchd

res

bhist

bjobs

bmod

bresize

bsub

 

7. Product Notifications

To receive information about product solution and patch updates automatically, subscribe to product notifications on the My notifications page ( www.ibm.com/support/mynotifications) on the IBM Support website (support.ibm.com). You can edit your subscription settings to choose the types of information you want to get notification about, for example, security bulletins, fixes, troubleshooting, and product enhancements or documentation changes.

 

8.   Copyright and trademark information

© Copyright IBM Corporation 2019 

U.S. Government Users Restricted Rights - Use, duplication or disclosure restricted by GSA ADP Schedule Contract with IBM Corp.

IBM®, the IBM logo and ibm.com® are trademarks of International Business Machines Corp., registered in many jurisdictions worldwide. Other product and service names might be trademarks of IBM or other companies. A current list of IBM trademarks is available on the Web at "Copyright and trademark information" at www.ibm.com/legal/copytrade.shtml.