CaGrid services deployment case study

From Informatics

Jump to: navigation, search

This page describes deployment of geWorkbench caGrid services on the development machine, afdev. The description here may be very specific or not the only way to get it work. It is more to highlight a successful case than to document various scenarios that the procedure failed. For more information, see Geworkbench Server Side.

Contents

Environment

The machine is afdev.c2b2.columbia.edu. The user account used for this purpose is cagrid.

Let us assume the basic environment has been set up. That is mainly caGrid 1.4, including globus. Cagrid's installer also has lead to install tomcat, and deploy its services to tomcat. Its instruction also specify the following environment variables: 

 ANT_HOME=
 CATALINA_HOME=
 JAVA_HOME=
 GLOBUS_LOCATION=

The tomcat instance we are using as of today is at /ifs/data/c2b2/af_lab/cagrid/apache-tomcat-5.5.27/, so CATALINA_HOME should be set as. I notice that this variable is not set in log-in script, please make sure to set it right before continuing.

  • From the cagrid install GUI, select the installation folder for Tomcat to be installed to as /ifs/data/c2b2/af_lab/cagrid

Prepare code and deploy

  • Prepare a directory called /ifs/data/c2b2/af_lab/cagrid/workspace under /ifs/data/c2b2/af_lab/cagrid
  • Download the code in /ifs/data/c2b2/af_lab/cagrid/workspace

Theoretically, running geWorkbench caGrid service install script is all we need to do (from workspace directory): 

cd /ifs/data/c2b2/af_lab/cagrid/workspace
ant -f geworkbench_server.xml install.

The following are more details that are probably useful:

  • The install target include two part: first checkout, then deploy. If you want to bypass the checkout part and only to re-deploy, run the deploy target only: 
ant -f geworkbench_server.xml install.

If you want to do checkout only: 

ant -f geworkbench_server.xml install.
  • The geWorkbench caGrid service install script covers the whole process. Sometimes it failed, I found by deleting all checked-out code and starting from scratch can fix the problem.
  • The script geworkbench_server.xml itself is part of the checkout. So int fact, you need to bootstrap by getting that script first if you don't have a updated copy already, e.g. when you start totally from scratch. You can do that by manually checking out the ginkgo module only and copy the script to the workspace directory. 
export CVSROOT=<username>@login.c2b2.columbia.edu:/cvs/magnet/cagrid
cvs checkout ginkgo
cp geworkbench_server.xml /ifs/data/c2b2/af_lab/cagrid/workspace

Please note that script will check out everything to /ifs/data/c2b2/af_lab/cagrid/workspace/cagrid_1_4

Start Tomcat and check the service status

Tomcat Commands

 > cd $CATALINA_HOME/bin
 > ./shutdown.sh
 > ./startup.sh

Check Tomcat Status

To report process status 

 > ps -ef\ grep tomcat

Display top CPU processes top [-] [d delay] [p pid] [q] [c] [C] [S] [s] [i] [n iter] [b 

 > top –i –c

Check Tomcat log files

 > cd $CATALINA_HOME/logs
 > tail catalina.out

Check Tomcat and Services Status

From a browser, check:

homepage http://afdev.c2b2.columbia.edu:8080/

index service http://afdev.c2b2.columbia.edu:8080/wsrf/services/DefaultIndexService

dispatcher http://afdev.c2b2.columbia.edu:8080/wsrf/services/cagrid/Dispatcher

individual service, e.g. Aracne http://afdev.c2b2.columbia.edu:8080/wsrf/services/cagrid/Aracne

Retrieved from "http://wiki.c2b2.columbia.edu/informatics/index.php/CaGrid_services_deployment_case_study"
Personal tools