The ${java.io.tmpdir} directory should be configurable from ./bin/jobscheduler_agent_instance.sh script

Current Situation

• When two or more Agent instances are running on the same server then the Agents read the same ./lib/log4j.xml configuration. 
• The log4j files has configuration ${java.io.tmpdir}/jade-client.log file, which creates the_ jade-client.log_ file at  ${java.io.tmpdir} directory.
• As a result, multiple instances of YADE JITL job are writing to the same log file.
• Also if a user wants to configure the log location or temp directory on a per Agent basis, it's not possible. According to KB article How do you change the directory JobScheduler uses for temporary files If a user tries to configure as JAVA_OPTIONS  JAVA_OPTIONS="-Djava.io.tmpdir=/tmp/agentA in ./bin/jobscheduler_agent_instance.sh  script, it does not have any effect, the Agent does not read this java.io.tmpdir parameter.

Desired Behavior

• The temp directory for should be configurable per agent basis in the  ./bin/jobscheduler_agent_instance.sh script.

Maintainer Note

• There are already all possibilities to assign Java options per Agent or per Job.
• Settings can be made in the instance script ./bin/jobscheduler_agent_[port].sh for each Agent instance separately.
• The environment variable JAVA_OPTIONS only determines the Java options of the Agent and not the Java options of the Jobs.
• If you set JAVA_OPTIONS="-Djava.io.tmpdir=/path/to/myDir" then you will see that the Agent also uses this temporary directory, because e.g. temporary files are created there by the Agent.
• Since JS-1751 the environment variable SCHEDULER_JOB_JAVA_OPTIONS was introduced in the Agent's instance script, with which the Java options of all jobs can be set which are processed by the Agent instance.
  • Here -Djava.io.tmpdir=/path/to/myDir has to be set to be considered by the logger of the JITL jobs such as YADE job.
  • You can also set Java options per Job with

    <job  process_class="myAgent" java_options="-Djava.io.tmpdir=/path/to/myDir">
        <params />
        <script  language="java:javascript">
            <![CDATA[
    function spooler_process(){
            spooler_log.info(java.lang.System.getProperty("java.io.tmpdir"));
    	return false;
    }
            ]]>
        </script>
        <run_time />
    </job>