8.11 Optimizing the Deployment of Reports

Before you deploy a report on a machine that is either slow or is running on a load, you may want to configure the following:

• Ping timeout (OPMN-side): Ping timeout is the measure that OPMN uses to determine the time that it must wait for a callback from an standalone Reports Server before considering it as a timeout.

  The default timeout period is 150. This period is calculated from: ping timeout, ping interval, and number of retries. The default values for these are:

  ping timeout = 30 seconds
  ping interval = 20 seconds
  number of retries = 3
  

  Note:

  The number of retries is applicable only when OPMN successfully connects to Standalone Reports Serverand receives regular ONS notifications from the process.

  Based on these values, there will be three ping attempts with a timeout of 30 seconds each at 20 second intervals. The first ping is done after the specified ping interval. Thus, from the time the Standalone Reports Server is started by OPMN, approximately 150 (20 + 3*30 + 2*20) seconds will elapse before the process is considered unresponsive and restarted. However, if after OPMN connects to Standalone Reports Server but server is too slow in sending regular ONS notifications, then the 30 second timeout applies.

  You can configure the ping timeout by adding a ping entry with sufficient timeout configured to the machine's load in following element in opmn.xml:

  <ias-component id="<reports_server_name>">
  ...
  ...
  <restart timeout="720" retry="2" />
  ...
  <ping timeout="110" interval="30" />
  ...
  

  Note:

  In 11g, Oracle WebLogic Server is not managed by OPMN, hence there will be no ping or death detection for In-Process Server running inside WLS_REPORTS.

• Reports Server start or restart timeout (OPMN.xml): Start or restart timeout is the measure that OPMN uses to determine the time that it must wait for Reports Server process type to start or restart (process-type id="ReportsServer" in opmn.xml) before considering it as a timeout.

  The default timeout period is 600. The default values for these are:

  <start timeout="600" retry="2"/>
  <restart timeout="600"/>
  

  When running on a loaded machine, an attempt to start all Reports Servers by OPMN may result in a start timeout for some Reports Servers as some of them were not able to finish the start up activities completely. Note that Reports Server also starts the number of engines specified in the initengine property of the engine element in the rwserver.conf file. Starting up these engine processes might take some time in loaded machines. In parallel to finetuning the Reports Server process start or restart property, you must also finetune the callbackTimeout property in rwserver.conf, as explained in the next bullet item.

• Callback timeout (Reports Server-side): Callback timeout is the measure that Reports Server uses to determine the time that it must wait for a response from the engine before timing out. You can specify this value in the rwserver.conf file. This time out period is in milliseconds.

  For example:

  <engine id="rwEng" class="oracle.reports.engine.EngineImpl" initEngine="1"   maxEngine="1" minEngine="0" engLife="50" maxIdle="30"    callbackTimeOut="90000">
  

  Note:

  Increase the callbackTimeOut value when the machine is very slow.