Hi,
We have similar case, as you have and but we are running Kepler 2.3.
Please find below what is our procedure, that works for our
applications/users.
I'm convinced others might give you better way to do it.
1. We are using central installation of the Kepler on the machine
2. We generate unique value for location of each instance of Kepler
running for each user
export HOME=$USER_HOME/kepler_tmp/${UNIQUE_VALUE}
export KEPLER_NEW_HOME=${HOME}/${KEPLER_SHADOW_DIR_NAME}
3. We do symbolic links to the central installation in that temporary
directory
4. We change configuration files to prevent mulitple kepler instances
from using the same hsql DB
(every single instance of kepler will be using separated hsql database
file)
(/common-2.3.0/resources/configurations/configuration.xml)
(/core-2.3.0/resources/configurations/configuration.xml )
5. We replace all necessary jars with provided kepler.jar with a patch
(there was one line needed to be patched in 2.3, I think it is not the
case for 2.4, but I'm not sure)
6. We run each instance of Kepler like that:
java -D"user.home"="${HOME}" -classpath
${KEPLER_NEW_HOME}/build-area/lib/ant.jar:${KEPLER_NEW_HOME}/kepler.jar
org.kepler.build.runner.Kepler -runwf -nogui ${PARAMETERS} -dir_path
${CURRENT_DIR}/ -nocache ${CURRENT_DIR}/${WORKFLOW_LOCATION} >&
${CURRENT_DIR}/kepler.log
I hope this will help you.
regards
Marcin Plociennik
--
Marcin Plociennik
Poznan Supercomputing and Networking Center
61-704 Poznan, ul. Noskowskiego 10 POLAND
phone (+48 61) 8582181
e-mail: marc...@man.poznan.pl
W dniu 2013-08-07 00:59, Lynch, Vickie E. pisze:
Hi,
We are developing a Kepler workflow for users that will run on a shared
computer for many hours/days optimizing models of experimental data. Our
requirement is for multiple users to run kepler workflows on a shared machine
and those users run multiple workflows. We have made the changes on this
webpage, https://kepler-project.org/developers/reference/accessing-hsql, so
that kepler will run for multiple users, but now each user can only run one
workflow at a time. Much time has been invested in developing a complicated
workflow and we did not find this problem until two developers started running
on the shared machine. It is not mentioned in the user's manual.
Is there some way to turn off the HSQL database? Will this problem be fixed in
a future release? Is there some workaround?
Vickie E. Lynch
Oak Ridge National Laboratory, PO Box 2008, Oak Ridge, TN 37831-6454
(865)250-4795, Bldg. 8600, Room C-509-F, MS6454, lync...@ornl.gov
_______________________________________________
Kepler-users mailing list
Kepler-users@kepler-project.org
http://lists.nceas.ucsb.edu/kepler/mailman/listinfo/kepler-users
_______________________________________________
Kepler-users mailing list
Kepler-users@kepler-project.org
http://lists.nceas.ucsb.edu/kepler/mailman/listinfo/kepler-users
