OpenbravoERP server customizing experience

Implementing OpenbravoERP in mid and big size companies, sometimes means having plenty of new developments and customizations, apart from having a bigger pool of users. Therefore the production server will need more resources.

When in a production environment there is a large server with plenty of resources, OpenbravoERP needs some parametrization in order to operate exploiting all resources properly.

Some improvements based on my experience, having a 20G server:

Related to Ant tasks:
In order to avoid java heap space when many sources have been developed, increase build.maxmemory in build.xml file. By default, from OB 2.50-MP12 ahead, this parameter is set to 1024M for 64bit servers and to 512M for 32bit servers. It will also compile faster.

Related to Tomcat:
In order to assure higher efficiency and lower response times to the users, customize some tomcat parameters. In file /etc/profile.d/tomcat.sh, change -Xmx parameter based in your own criteria. Realize 64bit servers need more resources than 32bit ones. The above is just an example:

export CATALINA_OPTS=”-server -Xms128M -Xmx2560M -XX:MaxPermSize=256M -Djava.library.path=/usr/lib64″

It is important to check the existing Tomcat documentation before you change anything, and change it first on a development environment.

Related to PostgreSQL:
In order to assure better database performance, edit /srv/pgsql/8.3/postgresql.conf and change this parameters shared_buffers, checkpoint_segments, maintenance_work_mem, wal_buffers and effective_cache_size.

Take into account that you will probably have to set a new value for SHMMAX, this can be done adding kernel.shmmax = 8589934592 to /etc/sysctl.conf. Where 8589934592 is the result of doing: 1024M (defined as shared_buffers) * 1024 bits/M * 8192.

Again, it is important to check the existing PostgreSQL documentation before you change anything, and change it first on a development environment.

Finally, it is important to execute vacuums frequently in order to have proper database statistics and performance. This can also be done using the console.

Remember, this is just my own experience.