IBM WebSphere Application Server Performance Cookbook

WebSphere Liberty requestTiming Recipe

WebSphere Liberty requestTiming gathers details on HTTP requests exceeding a configured threshold and generally has an overhead of less than 4%.

  1. The requestTiming-1.0 feature must be installed. Verify with productInfo; for example: 
    $ wlp/bin/productInfo featureInfo | grep requestTiming
requestTiming-1.0
    • If it's not installed, you may install it with installUtility and then restart Liberty; for example: 
      wlp/bin/installUtility install requestTiming-1.0
  2. Add the requestTiming-1.0 feature and a requestTiming element to server.xml (or a configDropin override) and change the attributes as needed. For example: 
    <featureManager><feature>requestTiming-1.0</feature></featureManager>
<requestTiming slowRequestThreshold="10s" hungRequestThreshold="10m" sampleRate="1" />
    1. slowRequestThreshold is the main setting and should generally be set to your maximum expected response time.
    2. hungRequestThreshold is not required and defaults to 10 minutes. It additionally gathers thread dumps when exceeded.
    3. Ideally, performance test requestTiming and increase sampleRate if testing shows an unacceptable overhead though this may cause missing some slow requests.
  3. If dynamic configuration updates are enabled (the default), then saving the configuration will enable requestTiming; otherwise, restart Liberty.
  4. Reproduce the problem and confirm the presence of TRAS0112W messages.
  5. Upload the following files:
    1. logs/messages*log
    2. logs/console.log
    3. logs/ffdc/*

For background, see WebSphere Liberty Request Timing.