Configuring Warmup Requests to Improve Performance

You can use warmup requests to reduce request and response latency during the time when your app's code is being loaded to a newly created instance.

App Engine frequently needs to load your app's code into a fresh instance. Loading an instance can happen in the following situations:

  • When you redeploy a version of your app.
  • When new instances are created due to the load from requests exceeding the capacity of the current set of running instances.
  • When maintenance and repairs of the underlying infrastructure or physical hardware occur.

Loading your app's code to a new instance can result in loading requests. Loading requests can result in increased request latency for your users, but you can avoid this latency using warmup requests. Warmup requests load your app's code into a new instance before any live requests reach that instance.

If warmup requests are enabled for your application, App Engine attempts to detect when your application needs a new instance and initiates a warmup request to initialize a new instance. However, these detection attempts do not work in every case. As a result, you might encounter loading requests, even if warmup requests are enabled in your app. For example, if your app is serving no traffic, the first request to the app will always be a loading request, not a warmup request.

Warmup requests use instance hours like any other request to your App Engine application. In most cases where warmup requests are enabled, you won't notice an increase in instance hours because your application is simply initializing in a warmup request instead of a loading request. Your instance hour usage can increase if you decide to do more work, such as pre-caching during a warmup request. If you set min_idle_instances to greater than 0, you might encounter warmup requests when those instances first start, but they will remain available after that time.

The default warmup request causes all JAR files to be indexed in memory and initializes your application and filters.

Enabling warmup requests

Warmup requests are used by the App Engine scheduler, which controls the auto scaling of instances based on user-supplied configuration. In the App Engine Java runtime, warmup requests are enabled by default, and so App Engine issues GET requests to /_ah/warmup, which allows you to respond and initialize your application's code as it requires. You can respond to warmup requests by using one of the following methods:

Using a <load-on-startup> servlet
The easiest way to provide warmup logic is to mark your own servlets as <load-on-startup> in the web.xml configuration file.
Using a ServletContextListener
Allows you to run custom logic before any of your servlets is first invoked either through a warmup request or a loading request.
Using a custom warmup servlet
Using a custom warmup servlet invokes the servlet's service method only during a warmup request rather than during loading requests.

You might need to implement your own handler for /_ah/warmup depending on which of these methods you choose.

Before you begin

When warmup requests are enabled, the scheduler starts up instances when it determines that more instances are needed. The scheduler uses warmup requests to start your app, so you'll see logs even if your app doesn't process those warmup requests.

The following message indicates warmup requests in the /_ah/warmup logs:

This request caused a new process to be started for your application, and thus caused your application code to be loaded for the first time. This request may thus take longer and use more CPU than a typical request for your application.

Note that warmup requests are not guaranteed to be called. In some situations loading requests are sent instead: for example, if the instance is the first one being started up, or if there is a steep ramp-up in traffic. However, there will be a "best effort" attempt to send requests to already warmed-up instances if warmup requests are enabled.

In Java 8, warmup requests are enabled by default. To enable them, add - warmup to the inbound_services directive in appengine-web.xml. Because by default, warmups are enabled, you only need to explicitly enable them if you have previously deployed an application with warmup requests disabled in the appengine-web.xml. If this is the case, you need to set the <warmup-requests-enabled> value to true and then redeploy.

Using a <load-on-startup> servlet

The easiest way to provide warmup logic is to mark your own servlets as <load-on-startup> in web.xml. This method requires no changes to your application code, and initializes all specified servlets when your application initializes.

In your web.xml file, for the servlets that you want to load on startup, add the <load-on-startup>1</load-on-startup> element to your <servlet> element. For example:


These lines load the specified servlet class and invoke the servlet's init() method. The warmup request initializes the specified servlets before servicing any live requests. However, if there is no warmup request, the servlets specified in <load-on-startup> are registered upon the first request to a new instance, which result in a loading request. As noted earlier, App Engine might not issue a warmup request every time your application needs a new instance.

Using a ServletContextListener

If you have custom logic that you want to run before any of your servlets is invoked:

  1. Register a ServletContextListener in your web.xml file.

  2. Supply a class alongside your servlet and filter code:

    public class MyListener implements ServletContextListener {
      public void contextInitialized(ServletContextEvent event) {
        // This will be invoked as part of a warmup request, or
        // the first user request if no warmup request was invoked.
      public void contextDestroyed(ServletContextEvent event) {
        // App Engine does not currently invoke this method.

The ServletContextListener runs during a warmup request. If there is no warmup request, it runs upon the first request to a new instance. This might result in loading requests.

Using a custom warmup servlet

The custom warmup servlet invokes the servlet's service method only during a warmup request. By placing expensive logic in a custom warmup servlet, you can avoid increased load times on loading requests.

To create a custom warmup servlet, simply override the built-in servlet definition for _ah_warmup in web.xml: