Document ID: 45305.1 Subject: Starting and stopping the Oracle Intelligent Agent Last Modified: 15 Apr 97 Author: ASHRIVES Starting the Intelligent Agent under OpenVMS ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ The Agent consists of two processes: the WORK process, which runs as a background detached process, and the COMM process, which runs as a subprocess of the WORK process. In addition, a third JOB process is spawned by the WORK process whenever needed to execute jobs. Steps: 1. Ensure the SQL*Net listener is already running. 2. The Agent is started up by LSNRCTL using the following command: $ LSNRCTL DBSNMP_START This command will create a detached process with a process name of the form ORA_TNS<number>, which will then spawn a subprocess. If a non-zero trace level is specified in SNMP.ORA, two trace files with names like DBSNMP_<pid>.TRC will be created in ORA_ROOT:[NETWORK.AGENT.TRACE]. The <pid> in these filenames correspond to those of the COMM and WORK processes. A trace file of the form DBSNMPJ_<pid>.TRC will also be created whenever a job is executed. Read the paragraph below on Agent Shutdown to understand the need for having GROUP privilege for the process starting up the Agent. NOTE: Allow a period of about 1-2 minutes for the creation of the processes and initialization of the Agent. This includes connection to the listener and database if specified in the visible services parameter of SNMP.ORA. It also includes initialization of the queue files. Checking if the agent is running ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ To check if the agent is running do: $ LSNRCTL DBSNMP_STATUS If the listener is running you will get a message saying the listener is already running. Stopping the Agent ~~~~~~~~~~~~~~~~~~ To stop the agent do: $ LSNRCTL DBSNMP_STOP Because this action depends on certain GROUP logicals, make sure that the account used to stop the Agent is in the same UIC group as the account that was used to start the Agent. GROUP privilege is also required to stop the agent with this command.