How can UGE use PostrgreSQL for the spooling Database?
For clusters running UGE, its plain text spool is rarely a performance bottleneck.
For busy enough clusters the BerkeleyDB is meant to be a faster way to spool, but it has stability issues at the time of writing (May 2015). A spool running on Postgres can be used instead, as follows:
1- Install the needed postgres packages, then initialize and start postgreSQL:
# yum install postgresql-server postgresql
# service postgresql initdb
# service postgresql start
# su - postgres
$ createuser -S -D -R -l -P -E ugeadmin
$ createdb -O ugeadmin ugespool
2- Edit "/cm/shared/apps/uge/var/cm/inst_template.conf" and change the following parameters:
PG_SPOOLING_ARGS="host=master dbname=ugespool user=ugeadmin"
The spooling arguments should be changed according to the Postgresql database configuration.
The cm-uge package provides the inst_template.conf.template file. This file should be copied to inst_template.conf so that it gets recognized by the UGE installer which gets called implicitly by the wlm-setup utility.
3- Set up UGE using wlm-setup
# wlm-setup -s -w uge -a <path to UGE tarballs>