SCore-D Test Procedure

Try the following test procedure on your server host to verify your cluster operation.
The bash(1) shell is used in the examples.
  1. Set the SCBDSERV and PATH environment variables

    If you have installed the system by your hand, please make sure that the SCBDSERV shell environment is set. If not, please login again. If you still do not see the variable, please make sure that a profile file for your login shell has been created under the /etc/profile.d, which is described in the server host settings section.

  2. Invoke the Compute Host Lock Client
    $ msgb -group pcc &
    pcc is the name of the group you defined in the scoreboard database.

  3. Compile and execute an MPC++ MTTL program

    Create the following program and call it hello.cc. A copy of the program is located in /opt/score/example/mttl/hello.cc: 
    #include <stdio.h>
#include <mpcxx.h>
main(int argc, char **argv) {
    mpcxx_spmd_initialize(argc, argv);
    printf("hello, world (from node %d)\n", myNode);
    exit(0);
}
    Compile it with mpc++:
    $ mpc++ -o hello hello.cc
    Execute the program locally: 
    $ scrun -nodes=1 ./hello
SCORE: connected (jid=100)
<0:0> SCORE: 1 node ready.
hello, world (from node 0)
$
    Execute the program on four nodes of the cluster: 
    $ scrun -nodes=4 ./hello
SCORE: connected (jid=100)
<0:0> SCORE: 4 nodes (4xq) ready.
hello, world (from node 2)
hello, world (from node 1)
hello, world (from node 3)
hello, world (from node 0)
$
  4. Compile and execute an MPICH-SCore program

    Copy the program /opt/score/example/mpi/cpi.c to your working directory and compile it with mpicc:
    $ mpicc -o cpi cpi.c -lm
    Execute the program locally using both scrun and mpirun: 
    $ scrun -nodes=1 ./cpi
SCORE: connected (jid=100)
<0:0> SCORE: 1 node ready.
Process 0 of 1 on comp3.score.rwcp.or.jp
pi is approximately 3.1416009869231254, Error is 0.0000083333333323
wall clock time = 0.000621
$ mpirun -np 1 ./cpi
SCORE: connected (jid=100)
<0:0> SCORE: 1 node ready.
Process 0 of 1 on comp3.score.rwcp.or.jp
pi is approximately 3.1416009869231254, Error is 0.0000083333333323
wall clock time = 0.000645
$
    Execute the program on four nodes of the cluster using both scrun and mpirun: 
    $ scrun -nodes=4 ./cpi
SCORE: connected (jid=100)
<0:0> SCORE: 4 nodes (4x1) ready.
Process 1 of 4 on comp1.score.rwcp.or.jp
Process 3 of 4 on comp3.score.rwcp.or.jp
Process 2 of 4 on comp2.score.rwcp.or.jp
Process 0 of 4 on comp0.score.rwcp.or.jp
pi is approximately 3.1416009869231245, Error is 0.0000083333333314
wall clock time = 0.000945
$ mpirun -np 4 ./cpi
SCORE: connected (jid=100)
<0:0> SCORE: 4 nodes (4x1) ready.
Process 2 of 4 on comp2.score.rwcp.or.jp
Process 1 of 4 on comp1.score.rwcp.or.jp
Process 0 of 4 on comp0.score.rwcp.or.jp
Process 3 of 4 on comp3.score.rwcp.or.jp
pi is approximately 3.1416009869231245, Error is 0.0000083333333314
wall clock time = 0.003627
$
  5. Exit the Single-User Environment
    $ exit
SCOUT: session done
$
  6. Start the SCore-D operating system for the Multi-User Environment

    Start scout and the SCore-D operating system as root. The startup of scored will take a few seconds to complete: 
    $ /bin/su -
# export SCBDSERV=`hostname`
# export PATH=$PATH:/opt/score/bin:/opt/score/sbin:/opt/score/deploy
# scout -g pcc
SCOUT: Spawn done.
SCOUT: session started
# scored
SYSLOG: Timeslice is set to 500[ms]
SYSLOG: Cluster[0]: comp0.score.rwcp.or.jp@0...comp3.score.rwcp.or.jp@3
SYSLOG:   BIN=linux, CPUGEN=pentium-iii, SMP=1, SPEED=500
SYSLOG:   Network[0]: myrinet/myrinet
SYSLOG: SCore-D network: myrinet/myrinet
SYSLOG: SCore-D server: comp3.score.rwcp.or.jp:9901
    You will see the node blocks in the msgb window change from blue to red.

  7. Execute an MPC++ MTTL program under the Multi-User Environment

    In a different shell, execute the program locally. You must specify the host where scored server is running. By default, this is the last host of the cluster group: 
    $ scrun -scored=comp3,nodes=1 ./hello
SCORE: connected (jid=100)
<0:0> SCORE: 1 node ready.
hello, world (from node 0)
$
    You will see output messages on the shell where scored was invoked, similar to the following: 
    SYSLOG: Login request: user1@host1.score.rwcp.or.jp:4556
SYSLOG: Login accepted: user1@host1.score.rwcp.or.jp:4556, JOB-ID: 100
  Nodes: 3-3, ./hello
SYSLOG: Logout: user1@host1.score.rwcp.or.jp:4556, JOB-ID: 100
    Execute the program on four nodes of the cluster group: 
    $ scrun -scored=comp3,nodes=4 ./hello
SCORE: connected (jid=200)
<0:0> SCORE: 4 nodes (4x1) ready.
hello, world (from node 2)
hello, world (from node 1)
hello, world (from node 3)
hello, world (from node 0)
$
  8. Execute an MPICH-SCore program under the Multi-User Environment

    Execute the program locally using both scrun and mpirun: 
    $ export SCORE_OPTIONS=scored=comp3
$ scrun -nodes=1 ./cpi
SCORE: connected (jid=300)
<0:0> SCORE: 1 node ready.
Process 0 of 1 on comp3.score.rwcp.or.jp
pi is approximately 3.1416009869231254, Error is 0.0000083333333323
wall clock time = 0.000621
$ mpirun -np 1 ./cpi
SCORE: connected (jid=400)
<0:0> SCORE: 1 node ready.
Process 0 of 1 on comp3.score.rwcp.or.jp
pi is approximately 3.1416009869231254, Error is 0.0000083333333323
wall clock time = 0.000645
$
    Execute the program on four nodes of the cluster using both scrun and mpirun: 
    $ scrun -nodes=4 ./cpi
SCORE: connected (jid=500)
<0:0> SCORE: 4 nodes (4x1) ready.
Process 1 of 4 on comp1.score.rwcp.or.jp
Process 3 of 4 on comp3.score.rwcp.or.jp
Process 2 of 4 on comp2.score.rwcp.or.jp
Process 0 of 4 on comp0.score.rwcp.or.jp
pi is approximately 3.1416009869231245, Error is 0.0000083333333314
wall clock time = 0.000945
$ mpirun -np 4 ./cpi
SCORE: connected (jid=600)
<0:0> SCORE: 4 nodes (4x1) ready.
Process 2 of 4 on comp2.score.rwcp.or.jp
Process 1 of 4 on comp1.score.rwcp.or.jp
Process 0 of 4 on comp0.score.rwcp.or.jp
Process 3 of 4 on comp3.score.rwcp.or.jp
pi is approximately 3.1416009869231245, Error is 0.0000083333333314
wall clock time = 0.003627
$
    If you wish to test more example programs you will find them located under the /opt/score/example directory.

  9. Stop the Multi-User Environment

    In a different window execute the following command to stop scored: 
    # sc_console comp3 -c shutdown
#
    You will see output messages on the shell where scored was invoked similar to the following: 
    SYSLOG: SCore-D shutdown.
SYSLOG: CONSOLE SHUTDOWN
#
    You will see the node blocks in the msgb window change back from red to blue.

    You can now exit from the scout session: 
    # exit
SCOUT: session done
#
Parallel and Distributed System Software Laboratory
Real World Computing Partnership
score-info@rwcp.or.jp

CREDIT
This document is a part of the SCore cluster system software developed at Real World Computing Partnership, Japan. Copyright (c) 2000, 1999 Real World Computing Partnership.