HP XC How To Installing Standard LSF on a Subset of XC Nodes Version 1.
© 2005 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. The only warranties for HP products and services are set forth in the express warranty statements accompanying such products and services. Nothing herein should be construed as constituting an additional warranty. HP shall not be liable for technical or editorial errors or omissions contained herein. Linux is a U.S. registered trademark of Linus Torvalds.
Contents Introduction Requirements ....................................................................................................................................... 5 Assumptions ........................................................................................................................................ 5 Sample Case ....................................................................................................................................... 5 Procedure Prepare the setup files..
Revision history Revision tables Table 1 Revisions Date Edition Revision July 2001 V1.
Introduction This document provides instructions for installing standard LSF on a subset of nodes in the XC cluster (in our example a set of large SMP nodes or "fat" nodes) while maintaining LSF-HPC integrated with SLURM on the rest of the nodes in the XC cluster (in our example the "thin" nodes).
Procedures Prepare the setup files 1. • Log into the head node of the XC cluster as root. Do not log in though the cluster alias. Change directory to /opt/hptc/lsf/top/conf and rename the existing setup files: # mv profile.lsf profile.lsf.xc # mv cshrc.lsf cshrc.lsf.xc Obtain standard LSF and install it into the existing LSF "tree" These instructions assume that the user is familiar with the procedures to install standard LSF, which basically consist of configuring the install.config file and running .
these files look for an XC-specific file: /etc/hptc-release. If found, these files assume that the LSF daemons should be interfacing with SLURM. For the fat nodes running standard LSF, this is not true. To work around this, we need to create an "identity" file that only exists on the thin nodes, and have the setup files for standard LSF look for this "identity" file. The "identity" file will be /var/lsf/lsfslurm. Use the pdsh command to create this file on all nodes except the fat nodes.
# source LSF . /opt/hptc/lsf/top/conf/profile.lsf.notxc # valid hosts for standard LSF on this cluster hosts="xc1 xc2 xc3 xc4 xc5 xc6" hostname=`hostname` valid=0 for i in $hosts; do if [ "$hostname" = "$i" ]; then valid=1 fi done if [ "$valid" = "0" ]; then exit 0 fi 2. 3. lsf_daemons "$1" Save and exit the file. Then set permissions, create the appropriate softlink, and enable it: # chmod 555 /opt/hptc/lsf/etc/slsf # ln -s /opt/hptc/lsf/etc/slsf /etc/init.
Re-run cluster_config to update node roles and re-image 1. 2. 3. 4. shutdown the rest of the cluster with stopsys. Change directory to /opt/hptc/config/sbin and execute ./cluster_config as follows: a. select "Modify Nodes" and change the roles on the fat nodes to remove the "compute" and "resource_management" roles. Ensure there's at least one "resource_management" role remaining in the cluster (2 "resource_management" nodes are recommended). b. do not re-install LSF. When .
<> <> xc1 $ bsub -I -n1 -R type=SLINUX64 hostname Job <177> is submitted to default queue . <> <> xc120 $ bsub -I -n2 -R type=SLINUX64 srun hostname Job <178> is submitted to default queue . <> <
