Trainings 2019 – Admin HPC – module 1 – sge installation

SGE installation

Summary

• Definition
• Prerequisites
• SGE installation
• Master Server installation
• Nodes installation
• Links
• License

Definition

SGE (SUN Grid Engine) is a linux job scheduler able to handle from 2 to thousands of servers at the same time.

One master server use the resources (CPU or RAM)of one or several nodes to perform analyses.

● An opensource tool

● 3 main functions :

• Allocates ressources (CPU,RAM) to users to allow them to
  launch their analyses

• Provides a frame to launch,execute et monitore the jobs on the
  whole allocated nodes

• Deals with jobs in queue wait

RPMS : https://copr.fedorainfracloud.org/coprs/loveshack/SGE/

SOURCE : https://arc.liv.ac.uk/downloads/SGE/releases/

Prerequisites:

you have to choose a nfs parttion shared between your master and your node.

$yum -y install epel-release
$ yum -y install jemalloc-devel openssl-devel ncurses-devel pam-devel libXmu-devel hwloc-devel hwloc hwloc-libs java-devel javacc ant-junit libdb-devel motif-devel csh ksh xterm db4-utils perl-XML-Simple perl-Env xorg-x11-fonts- ISO8859-1-100dpi xorg-x11-fonts-ISO8859-1-75dpi
$ yum install –y gcc
$ groupadd -g 490 sgeadmin
$ useradd -u 495 -g 490 -r -m  -d /home/sgeadmin -s /bin/bash sgeadmin

SGE installation:

 $ wget https://arc.liv.ac.uk/downloads/SGE/releases/8.1.9/sge-8.1.9.tar.gz
 $ cd sge-8.1.9/source/
 $ sh scripts/bootstrap.sh && ./aimk && ./aimk –man
 $ export SGE_ROOT=<NFS_SHARE_DIR> && mkdir –p $SGE_ROOT
 $ echo Y | ./scripts/distinst -local -allall -libs –noexit
 $ chown -R sgeadmin.sgeadmin $SGE_ROOT

Master server installation:

On the master server type:

$ yum -y install nfs-utils
$ cd $SGE_ROOT
$ ./install_qmaster

You will be asked to mention:

• The Grid Engine administrator name

• The value of $SGE_ROOT

• The number of the TCP port ( 6444 by default)

• The Grid Engine service Name

• The number of the TCP port for sge_execd (6445 by default)

• The Grid Engine cell name (leave default)

• The spool directory path

• The spooling method

• group ID range for Grid Engine

• Accept the installation of booting scripts

• Do not install shadow server

• The list of administration and submit hosts : put your master server in both

To finish the installation, launch the following commands:

$  cp $SGE_ROOT/default/common/settings.sh /etc/profile.d/SGE.sh
$ cp $SGE_ROOT/default/common/settings.csh /etc/profile.d/SGE.csh

If you have serveral nodes: allow the nfs export of the SGE PATH to the other nodes of the cluster:

In /etc/exports add:

$ /usr/local/sge  IP_range_nodes/24(rw,no_root_squash)

Start and enable the nfs service:

$ systemctl start rpcbind nfs-server
$ systemctl enable rpcbind nfs-server

Nodes installation:

if you have several nodes, make sure they appear in the /etc/hosts file

Mount the SGE_ROOT path on the node:

 $ yum -y install nfs-utils
 $ systemctl start rpcbind
 $ systemctl enable rpcbind
 $ mount –t nfs <master>:/usr/local/sge /usr/local/sge
 $ systemctl start rpcbind nfs-server
 $ systemctl enable rpcbind nfs-server

Launch the installation of the node:

 $ export SGE_ROOT=/usr/local/sge
 $ export SGE_CELL=default
 $ cd $SGE_ROOT
 $ ./install_execd 

After answering all the questions:

$ cp $SGE_ROOT/default/common/settings.sh /etc/profile.d/SGE.sh
$ cp $SGE_ROOT/default/common/settings.csh /etc/profile.d/SGE.csh

Links

• Related courses : [HPC Trainings]((https://itrop.ird.fr/wordpress/index.php/trainings-2019-hpc/)

License

The resource material is licensed under the Creative Commons Attribution 4.0 International License (here).