Beginning with version 2.3, Opal provides an ability to queue up jobs using the Fork Job Manager. This feature may be particularly useful if the jobs being launched by Opal need a definite amount of memory or processing, thus hindering an large amount of jobs from running at the same time. Note that this is not a replacement for a traditional batch scheduler - it is simply a way to limit the total number of jobs that are concurrently being executed. Also note that we have not done any scalability testing of this feature because the Fork Job Manager is not supposed to be used for large numnbers of concurrent jobs.
Users can set appropriate properties in the etc/opal.properties to configure this features. For instance, the specific lines for configuring this feature are as follows:
## BEGIN: information for Fork job manager ## ---------------------------------------------------------- # number of jobs that can be in execution simultaneously fork.jobs.limit=1 ## ---------------------------------------------------------- ## END: information for For job manager |
By default, the job limits are turned off. To turn it on, please set the fork.jobs.limit to the number of jobs that can execute concurrently.
As usual, after all changes to the etc/opal.properties, reinstall Opal using the ant install command, and restart Tomcat for the changes to take effect.