Test CREAM
Direct submission
UI
Ref:
CREAM-Guide
Cream client configuration
Log files will be created only if command executed with "-d" flag (debugging)
Environment variable GLITE_CREAM_CLIENT_CONFIG will be read first, if cream client config file is not specified with executed command.
Cream UI config example: /opt/glite/etc/glite-ce-cream-cli/glite_cream.conf.template
[kar@ui cream]$ cat $GLITE_CREAM_CLIENT_CONFIG
[
STATUS_VERBOSITY_LEVEL = 2;
SUBMIT_LOG_DIR="/home/kar/cream/logs";
DELEGATE_LOG_DIR="/home/kar/cream/logs";
STATUS_LOG_DIR="/home/kar/cream/logs";
SUSPEND_LOG_DIR="/home/kar/cream/logs";
RESUME_LOG_DIR="/home/kar/cream/logs";
CANCEL_LOG_DIR="/home/kar/cream/logs";
PURGE_LOG_DIR="/home/kar/cream/logs";
PROXYRENEW_LOG_DIR="/home/kar/cream/logs";
]
JDL_DEFAULT_ATTRIBUTES variables can be specified in ui cream config.
[
JDL_DEFAULT_ATTRIBUTES = [
JobType="Normal";
Type="Job";
];
STATUS_VERBOSITY_LEVEL = 2;
SUBMIT_LOG_DIR="/home/kar/cream/logs";
DELEGATE_LOG_DIR="/home/kar/cream/logs";
STATUS_LOG_DIR="/home/kar/cream/logs";
SUSPEND_LOG_DIR="/home/kar/cream/logs";
RESUME_LOG_DIR="/home/kar/cream/logs";
CANCEL_LOG_DIR="/home/kar/cream/logs";
PURGE_LOG_DIR="/home/kar/cream/logs";
PROXYRENEW_LOG_DIR="/home/kar/cream/logs";
]
Ref:
CREAM-JDL
Looking for creamCE
[kar@ui cream]$ lcg-infosites --vo balticgrid ce --is bdii.grid.vgtu.lt|grep cream
4 4 0 0 0 ce1.grid.vgtu.lt:8443/cream-pbs-balticgrid
Job Types
Normal
Parametric
MPICH
Check what
MPI implementations cluster supports
[kar@ui cream]$ ldapsearch -LLL -x -h ce1.grid.vgtu.lt:2170 -b mds-vo-name=VGTU-CREAM,o=grid|egrep "MPI-START|MPI_SHARED_HOME|MPICH|OPENMPI"
GlueHostApplicationSoftwareRunTimeEnvironment: MPICH
GlueHostApplicationSoftwareRunTimeEnvironment: MPICH-1.2.7p1
GlueHostApplicationSoftwareRunTimeEnvironment: MPICH2
GlueHostApplicationSoftwareRunTimeEnvironment: MPICH2-1.0.4
GlueHostApplicationSoftwareRunTimeEnvironment: OPENMPI
GlueHostApplicationSoftwareRunTimeEnvironment: OPENMPI-1.2.7
GlueHostApplicationSoftwareRunTimeEnvironment: MPI-START
GlueHostApplicationSoftwareRunTimeEnvironment: MPI_SHARED_HOME
Check if submission to CREAM-CE is allowed:
[kar@ui cream]$ glite-ce-allowed-submission ce1.grid.vgtu.lt
Job Submission to this CREAM CE is enabled
Prepare job input files:
[kar@ui mpich]$ export DPNS_HOST=se.grid.vgtu.lt
[kar@ui mpich]$ dpns-mkdir -p /dpm/grid.vgtu.lt/home/balticgrid/kar
[kar@ui mpich]$ lcg-cp file://$PWD/mpi-hooks.sh srm://se.grid.vgtu.lt/dpm/grid.vgtu.lt/home/balticgrid/kar/mpi-hooks.sh
[kar@ui mpich]$ lcg-cp file://$PWD/mpi-start-wrapper.sh srm://se.grid.vgtu.lt/dpm/grid.vgtu.lt/home/balticgrid/kar/mpi-start-wrapper.sh
[kar@ui mpich]$ lcg-cp file://$PWD/pingtest.c srm://se.grid.vgtu.lt/dpm/grid.vgtu.lt/home/balticgrid/kar/pingtest.c
[kar@ui mpich]$ dpns-ls /dpm/grid.vgtu.lt/home/balticgrid/kar
mpi-hooks.sh
mpi-start-wrapper.sh
pingtest.c
Set variable for cream ui config:
[kar@ui mpich]$ export GLITE_CREAM_CLIENT_CONFIG=/home/kar/cream/ui_cream.conf
Prepare JDL file:
[kar@ui mpich]$ cat mpi-start.jdl
[
CpuNumber = 4;
Executable = "mpi-start-wrapper.sh";
Arguments = "pingtest MPICH";
StdOutput = "mpi-start.out";
StdError = "mpi-start.err";
InputSandbox = {
"gsiftp://se.grid.vgtu.lt/dpm/grid.vgtu.lt/home/balticgrid/kar/pingtest.c",
"gsiftp://se.grid.vgtu.lt/dpm/grid.vgtu.lt/home/balticgrid/kar/mpi-start-wrapper.sh",
"gsiftp://se.grid.vgtu.lt/dpm/grid.vgtu.lt/home/balticgrid/kar/mpi-hooks.sh"
};
OutputSandbox = {
"mpi-start.err",
"mpi-start.out"
};
OutputSandboxBaseDestURI = "gsiftp://se.grid.vgtu.lt/dpm/grid.vgtu.lt/home/balticgrid/kar";
]
Submit Job:
[kar@ui mpich]$ glite-ce-job-submit -D 2 -r ce1.grid.vgtu.lt/cream-pbs-balticgrid mpi-start.jdl
https://ce1.grid.vgtu.lt:8443/CREAM605721465
Job's status:
[kar@ui mpich]$ glite-ce-job-status https://ce1.grid.vgtu.lt:8443/CREAM605721465
****** JobID=[https://ce1.grid.vgtu.lt:8443/CREAM605721465]
Current Status = [DONE-OK]
Working Dir = [[reserved]]
ExitCode = [0]
Grid JobID = [N/A]
LRMS Abs JobID = [[reserved]]
LRMS JobID = [[reserved]]
Deleg Proxy ID = [2]
DelegProxyInfo = [Valid From : 11/23/08 6:09 PM
Valid To : 11/24/08 6:10 AM
Holder Subject : DC=org,DC=balticgrid,OU=vgtu.lt,CN=Karolis Eigelis
Holder CA : DC=org,DC=balticgrid,CN=Baltic Grid Certification Authority
VO : balticgrid
AC Issuer : CN=host/voms.balticgrid.org, OU=eenet.ee, DC=balticgrid, DC=org
Attribute : /balticgrid/Role=NULL/Capability=NULL
]
Worker Node = [compute-0-1.local]
Local User = [bg17]
CREAM ISB URI = [gsiftp://ce1.grid.vgtu.lt/opt/glite/var/cream_sandbox/balticgrid/DC_org_DC_balticgrid_OU_vgtu.lt_CN_Karolis_Eigelis_balticgrid_Role_NULL_Capability_NULL/CREAM605721465/ISB]
CREAM OSB URI = [gsiftp://ce1.grid.vgtu.lt/opt/glite/var/cream_sandbox/balticgrid/DC_org_DC_balticgrid_OU_vgtu.lt_CN_Karolis_Eigelis_balticgrid_Role_NULL_Capability_NULL/CREAM605721465/OSB]
JDL = [[ Arguments = "pingtest MPICH"; OutputSandboxBaseDestURI = "gsiftp://se.grid.vgtu.lt/dpm/grid.vgtu.lt/home/balticgrid/kar"; QueueName = "balticgrid"; Executable = "mpi-start-wrapper.sh"; CpuNumber = 4; StdOutput = "mpi-start.out"; OutputSandbox = { "mpi-start.err","mpi-start.out" }; VirtualOrganisation = "balticgrid"; StdError = "mpi-start.err"; BatchSystem = "pbs"; InputSandbox = { "gsiftp://se.grid.vgtu.lt/dpm/grid.vgtu.lt/home/balticgrid/kar/pingtest.c","gsiftp://se.grid.vgtu.lt/dpm/grid.vgtu.lt/home/balticgrid/kar/mpi-start-wrapper.sh","gsiftp://se.grid.vgtu.lt/dpm/grid.vgtu.lt/home/balticgrid/kar/mpi-hooks.sh" } ]]
Type = [normal]
Job status changes:
-------------------
Status = [REGISTERED] - [Sun 23 Nov 2008 23:19:17] (1227475157)
Status = [PENDING] - [Sun 23 Nov 2008 23:19:17] (1227475157)
Status = [RUNNING] - [Sun 23 Nov 2008 23:19:23] (1227475163)
Status = [REALLY-RUNNING] - [Sun 23 Nov 2008 23:19:29] (1227475169)
Status = [DONE-OK] - [Sun 23 Nov 2008 23:19:50] (1227475190)
Issued Commands:
-------------------
*** Command Name = [JOB_REGISTER]
Command Category = [JOB_MANAGEMENT]
Command Status = [SUCCESSFULL]
Creation Time = [Sun 23 Nov 2008 23:19:17] (1227475157)
Start Scheduling Time = [Sun 23 Nov 2008 23:19:17] (1227475157)
Start Processing Time = [Sun 23 Nov 2008 23:19:17] (1227475157)
Execution Completed Time = [Sun 23 Nov 2008 23:19:17] (1227475157)
*** Command Name = [JOB_START]
Command Category = [JOB_MANAGEMENT]
Command Status = [SUCCESSFULL]
Creation Time = [Sun 23 Nov 2008 23:19:17] (1227475157)
Start Scheduling Time = [Sun 23 Nov 2008 23:19:17] (1227475157)
Start Processing Time = [Sun 23 Nov 2008 23:19:17] (1227475157)
Execution Completed Time = [Sun 23 Nov 2008 23:19:25] (1227475165)
*** Command Name = [PROXY_RENEW]
Command Category = [JOB_MANAGEMENT]
Command Status = [SUCCESSFULL]
Creation Time = [Sun 23 Nov 2008 23:19:29] (1227475169)
Start Scheduling Time = [Sun 23 Nov 2008 23:19:29] (1227475169)
Start Processing Time = [Sun 23 Nov 2008 23:19:29] (1227475169)
Execution Completed Time = [Sun 23 Nov 2008 23:19:36] (1227475176)
Get Job's output:
[kar@ui mpich]$ lcg-ls srm://se.grid.vgtu.lt/dpm/grid.vgtu.lt/home/balticgrid/kar
/dpm/grid.vgtu.lt/home/balticgrid/kar/mpi-hooks.sh
/dpm/grid.vgtu.lt/home/balticgrid/kar/mpi-start-wrapper.sh
/dpm/grid.vgtu.lt/home/balticgrid/kar/mpi-start.err
/dpm/grid.vgtu.lt/home/balticgrid/kar/mpi-start.out
/dpm/grid.vgtu.lt/home/balticgrid/kar/pingtest.c
[kar@ui mpich]$ lcg-cp srm://se.grid.vgtu.lt/dpm/grid.vgtu.lt/home/balticgrid/kar/mpi-start.err file://$PWD/mpi-start.err
[kar@ui mpich]$ lcg-cp srm://se.grid.vgtu.lt/dpm/grid.vgtu.lt/home/balticgrid/kar/mpi-start.out file://$PWD/mpi-start.out
[kar@ui mpich]$ cat mpi-start.out
Compiling /home/bg17/home_cream_189927535/CREAM189927535/pingtest
mpicc -o /home/bg17/home_cream_189927535/CREAM189927535/pingtest /home/bg17/home_cream_189927535/CREAM189927535/pingtest.c
Successfully compiled /home/bg17/home_cream_189927535/CREAM189927535/pingtest
message size transfertime bandwidth
32 bytes 0.000029 sec 1.109570 MB/s
2048 bytes 0.000031 sec 66.884389 MB/s
131072 bytes 0.000858 sec 152.696941 MB/s
8388608 bytes 0.115781 sec 72.452083 MB/s
Executing post hook.
Nodes:
compute-0-1.local
compute-0-1.local
compute-0-0.local
compute-0-0.local
Finished the post hook.
glite-ce-allowed-submission
Check if submission to CREAM-CE is allowed:
[kar@ui cream]$ glite-ce-allowed-submission ce1.grid.vgtu.lt
Job Submission to this CREAM CE is enabled
glite-ce-delegate-proxy
[kar@ui cream]$ voms-proxy-init --voms balticgrid
Enter GRID pass phrase:
Your identity: /DC=org/DC=balticgrid/OU=vgtu.lt/CN=Karolis Eigelis
Creating temporary proxy .................................................................. Done
Contacting voms.balticgrid.org:15000 [/DC=org/DC=balticgrid/OU=eenet.ee/CN=host/voms.balticgrid.org] "balticgrid" Done
Creating proxy ................................. Done
Your proxy is valid until Mon Nov 24 08:10:57 2008
[kar@ui cream]$ glite-ce-delegate-proxy -e ce1.grid.vgtu.lt:8443 2
2008-11-23 20:13:29,855 NOTICE - Proxy with delegation id [2] succesfully delegated to endpoint [https://ce1.grid.vgtu.lt:8443//ce-cream/services/gridsite-delegation]
glite-ce-proxy-renew
[kar@ui cream]$ glite-ce-proxy-renew -e ce1.grid.vgtu.lt 2
2008-11-23 20:14:50,345 NOTICE - Proxy with delegation id [2] succesfully renewed to endpoint [https://ce1.grid.vgtu.lt:8443//ce-cream/services/gridsite-delegation]
glite-ce-job-submit
[kar@ui cream]$ glite-ce-job-submit -D 2 -r ce1.grid.vgtu.lt/cream-pbs-balticgrid mpi-start.jdl
https://ce1.grid.vgtu.lt:8443/CREAM324798604
glite-ce-job-status
[kar@ui mpich]$ glite-ce-job-status https://ce1.grid.vgtu.lt:8443/CREAM605721465
****** JobID=[https://ce1.grid.vgtu.lt:8443/CREAM605721465]
Current Status = [DONE-OK]
Working Dir = [[reserved]]
ExitCode = [0]
Grid JobID = [N/A]
LRMS Abs JobID = [[reserved]]
LRMS JobID = [[reserved]]
Deleg Proxy ID = [2]
DelegProxyInfo = [Valid From : 11/23/08 6:09 PM
Valid To : 11/24/08 6:10 AM
Holder Subject : DC=org,DC=balticgrid,OU=vgtu.lt,CN=Karolis Eigelis
Holder CA : DC=org,DC=balticgrid,CN=Baltic Grid Certification Authority
VO : balticgrid
AC Issuer : CN=host/voms.balticgrid.org, OU=eenet.ee, DC=balticgrid, DC=org
Attribute : /balticgrid/Role=NULL/Capability=NULL
]
Worker Node = [compute-0-1.local]
Local User = [bg17]
CREAM ISB URI = [gsiftp://ce1.grid.vgtu.lt/opt/glite/var/cream_sandbox/balticgrid/DC_org_DC_balticgrid_OU_vgtu.lt_CN_Karolis_Eigelis_balticgrid_Role_NULL_Capability_NULL/CREAM605721465/ISB]
CREAM OSB URI = [gsiftp://ce1.grid.vgtu.lt/opt/glite/var/cream_sandbox/balticgrid/DC_org_DC_balticgrid_OU_vgtu.lt_CN_Karolis_Eigelis_balticgrid_Role_NULL_Capability_NULL/CREAM605721465/OSB]
JDL = [[ Arguments = "pingtest MPICH"; OutputSandboxBaseDestURI = "gsiftp://se.grid.vgtu.lt/dpm/grid.vgtu.lt/home/balticgrid/kar"; QueueName = "balticgrid"; Executable = "mpi-start-wrapper.sh"; CpuNumber = 4; StdOutput = "mpi-start.out"; OutputSandbox = { "mpi-start.err","mpi-start.out" }; VirtualOrganisation = "balticgrid"; StdError = "mpi-start.err"; BatchSystem = "pbs"; InputSandbox = { "gsiftp://se.grid.vgtu.lt/dpm/grid.vgtu.lt/home/balticgrid/kar/pingtest.c","gsiftp://se.grid.vgtu.lt/dpm/grid.vgtu.lt/home/balticgrid/kar/mpi-start-wrapper.sh","gsiftp://se.grid.vgtu.lt/dpm/grid.vgtu.lt/home/balticgrid/kar/mpi-hooks.sh" } ]]
Type = [normal]
Job status changes:
-------------------
Status = [REGISTERED] - [Sun 23 Nov 2008 23:19:17] (1227475157)
Status = [PENDING] - [Sun 23 Nov 2008 23:19:17] (1227475157)
Status = [RUNNING] - [Sun 23 Nov 2008 23:19:23] (1227475163)
Status = [REALLY-RUNNING] - [Sun 23 Nov 2008 23:19:29] (1227475169)
Status = [DONE-OK] - [Sun 23 Nov 2008 23:19:50] (1227475190)
Issued Commands:
-------------------
*** Command Name = [JOB_REGISTER]
Command Category = [JOB_MANAGEMENT]
Command Status = [SUCCESSFULL]
Creation Time = [Sun 23 Nov 2008 23:19:17] (1227475157)
Start Scheduling Time = [Sun 23 Nov 2008 23:19:17] (1227475157)
Start Processing Time = [Sun 23 Nov 2008 23:19:17] (1227475157)
Execution Completed Time = [Sun 23 Nov 2008 23:19:17] (1227475157)
*** Command Name = [JOB_START]
Command Category = [JOB_MANAGEMENT]
Command Status = [SUCCESSFULL]
Creation Time = [Sun 23 Nov 2008 23:19:17] (1227475157)
Start Scheduling Time = [Sun 23 Nov 2008 23:19:17] (1227475157)
Start Processing Time = [Sun 23 Nov 2008 23:19:17] (1227475157)
Execution Completed Time = [Sun 23 Nov 2008 23:19:25] (1227475165)
*** Command Name = [PROXY_RENEW]
Command Category = [JOB_MANAGEMENT]
Command Status = [SUCCESSFULL]
Creation Time = [Sun 23 Nov 2008 23:19:29] (1227475169)
Start Scheduling Time = [Sun 23 Nov 2008 23:19:29] (1227475169)
Start Processing Time = [Sun 23 Nov 2008 23:19:29] (1227475169)
Execution Completed Time = [Sun 23 Nov 2008 23:19:36] (1227475176)
glite-ce-job-suspend
Possible only on these job states:
glite-ce-job-resume
Possible only on this job state:
glite-ce-job-cancel
Possible only on these job states:
- Pending
- Idle
- Running
- Really-Running
- Held
glite-ce-job-purge
Possible only on these job states:
- Registered
- Done-OK
- Done-Failed
- Aborted
- Cancelled
glite-ce-job-list
[kar@ui cream]$ glite-ce-job-list ce1.grid.vgtu.lt
https://ce1.grid.vgtu.lt:8443/CREAM064670125
https://ce1.grid.vgtu.lt:8443/CREAM070084752
https://ce1.grid.vgtu.lt:8443/CREAM122863713
https://ce1.grid.vgtu.lt:8443/CREAM147794363
https://ce1.grid.vgtu.lt:8443/CREAM168366289
https://ce1.grid.vgtu.lt:8443/CREAM183387815
https://ce1.grid.vgtu.lt:8443/CREAM194272022
https://ce1.grid.vgtu.lt:8443/CREAM255920954
https://ce1.grid.vgtu.lt:8443/CREAM260357786
https://ce1.grid.vgtu.lt:8443/CREAM282830659
creamCE notes
CE configured with YAIM
/opt/glite/yaim/bin/yaim -c -s site-info-VGTU-CREAM.def -n MPI_CE -n creamCE -n TORQUE_server -n TORQUE_utils -n MON -n BDII_site
WN configured with YAIM
/opt/glite/yaim/bin/yaim -c -s site-info-VGTU-CREAM.def -n MPI_WN -n glite-WN -n TORQUE_client -n TORQUE_utils
Torque server cosmetics to run
MPI jobs on multi-cores; N = total number of CPUs
qmgr -c 'set server resources_available.nodect = N'
MPI-START
site-info.def for default mpi-start:
MPI_MPICH_ENABLE="yes"
MPI_MPICH2_ENABLE="yes"
MPI_OPENMPI_ENABLE="yes"
MPI_MPICH_PATH="/opt/mpich-1.2.7p1/"
MPI_MPICH_VERSION="1.2.7p1"
MPI_MPICH2_PATH="/opt/mpich2-1.0.4/"
MPI_MPICH2_VERSION="1.0.4"
MPI_OPENMPI_VERSION="1.2.7"
MPI_OPENMPI_PATH="/usr/lib/openmpi/1.2.7-gcc"
MPI_SHARED_HOME="yes"
MPI_SSH_HOST_BASED_AUTH="no"
More info about
MPI-START
here.
--
KarolisEigelis - 19 Nov 2008