Hi,
Oracle Sungrid engine is mainly used for clustering, in which user can submit jobs on the machines on the cluster.
There will be a master node , submit node and then slave machines in which jobs are submitted.
Below is the steps to install sungrid engine client on the machines. In this scenario master is already installed on the cluster. Here we will install and configure a machine as slave on that cluster.
yum -y install binutils csh java ksh libXpm openmotif22 tcl
yum --enablerepo=epel install gridengine gridengine-qmaster gridengine-execd gridengine-qmon gridengine-devel
cd /usr/share/gridengine/
./inst_sge -x -m
just proceed with the installation.
When the grid engine is installed , there will be two directories in which grid
engine data is found.
One is /usr/share/gridengine = $SGE_ROOT
in which all configuration files resides
another configuration root is $SGE_CELL which is named as default/common directory
inside the $SGE_ROOT derectory
source /usr/share/gridengine/default/common/settings.sh
chkconfig --level 35 sge_execd on
chkconfig --level 35 sgemaster on
qconf -as `hostname`
qconf -msconf
Now you have to make sure that the machine must need a hostname. Set it in
/etc/sysconfig/network.Then add the hostname in DNS so that hostname is pinging
from any where especially from the master node.
now go to
cd /usr/share/gridengine/defaults/common
Edit the configuration file act_qmaster.
vi act_qmaster
add the hostmame of the master node here.
suppose say the hostname of master kmaster.example.com
and hostname of the new node we are creating is knode01.example.com
Now ssh into the master node
kmaster.example.com
run the command qconf -shgrpl
this command will show the list of available host ques.
the queues will be lised as
@queue1
@queue2
etc
now you need to add the new machine to any of the queue on the master node.
using the command qconf -mhgrp @queue1
a new text file will open and you need to add the hostname of the new machine
(knode01.example.com) to it
save the file in VI.
Now you need to add the machine as submit host and administrative host to run the
jobs in them
so use the command
qconf -ah knode01.example.com
qconf -as knode01.example.com
now exit from the master node and ssh again into the new slave node.
go to directory /usr/share/gridengine
run the command ./install_execd
type in yes when prompted.
This will install the slave node and will add it to the master node.
after the installation is finished.
Please start the service /etc/init,d/sge_execd
and then run the command qstat -f
if every thing is correct , this command will list all the available nodes
on the grid.
Regards
Syamkumar.M
Oracle Sungrid engine is mainly used for clustering, in which user can submit jobs on the machines on the cluster.
There will be a master node , submit node and then slave machines in which jobs are submitted.
Below is the steps to install sungrid engine client on the machines. In this scenario master is already installed on the cluster. Here we will install and configure a machine as slave on that cluster.
yum -y install binutils csh java ksh libXpm openmotif22 tcl
yum --enablerepo=epel install gridengine gridengine-qmaster gridengine-execd gridengine-qmon gridengine-devel
cd /usr/share/gridengine/
./inst_sge -x -m
just proceed with the installation.
When the grid engine is installed , there will be two directories in which grid
engine data is found.
One is /usr/share/gridengine = $SGE_ROOT
in which all configuration files resides
another configuration root is $SGE_CELL which is named as default/common directory
inside the $SGE_ROOT derectory
source /usr/share/gridengine/default/common/settings.sh
chkconfig --level 35 sge_execd on
chkconfig --level 35 sgemaster on
qconf -as `hostname`
qconf -msconf
Now you have to make sure that the machine must need a hostname. Set it in
/etc/sysconfig/network.Then add the hostname in DNS so that hostname is pinging
from any where especially from the master node.
now go to
cd /usr/share/gridengine/defaults/common
Edit the configuration file act_qmaster.
vi act_qmaster
add the hostmame of the master node here.
suppose say the hostname of master kmaster.example.com
and hostname of the new node we are creating is knode01.example.com
Now ssh into the master node
kmaster.example.com
run the command qconf -shgrpl
this command will show the list of available host ques.
the queues will be lised as
@queue1
@queue2
etc
now you need to add the new machine to any of the queue on the master node.
using the command qconf -mhgrp @queue1
a new text file will open and you need to add the hostname of the new machine
(knode01.example.com) to it
save the file in VI.
Now you need to add the machine as submit host and administrative host to run the
jobs in them
so use the command
qconf -ah knode01.example.com
qconf -as knode01.example.com
now exit from the master node and ssh again into the new slave node.
go to directory /usr/share/gridengine
run the command ./install_execd
type in yes when prompted.
This will install the slave node and will add it to the master node.
after the installation is finished.
Please start the service /etc/init,d/sge_execd
and then run the command qstat -f
if every thing is correct , this command will list all the available nodes
on the grid.
Regards
Syamkumar.M
