如何在JMeter中实现多用户共享HTTP缓存管理器缓存？

Hey there, I totally get what you're trying to do—you want a single cache instance that all 10 of your test threads can reuse, instead of each thread maintaining its own separate cache. The default HTTP Cache Manager in JMeter is thread-isolated by design, even when placed outside the Thread Group, so we need to work around that. Here are three reliable approaches to make global cache sharing work:

1. Use the HTTP Cache Manager (Global) Plugin #

This is the simplest and most straightforward solution, purpose-built for cross-thread cache sharing.

• First, install the JMeter Plugins Manager if you haven't already.
• Open the Plugins Manager, search for HTTP Cache Manager (Global), and install it.
• Remove your existing HTTP Cache Manager, then add the HTTP Cache Manager (Global) to the root of your Test Plan (outside all Thread Groups).
• Configure it just like you would the regular cache manager—set cache policies, exclude/include URLs, etc. Now all threads will share this single global cache pool.

2. Manually Implement Global Cache with JSR223 Scripts #

If you prefer not to use plugins, you can build a custom global cache using JMeter's shared properties and JSR223 elements:

• Step 1: Populate the global cache
  Add a setUp Thread Group (this runs once before your main test threads) and put your cache-warming HTTP request inside it. Then add a JSR223 PostProcessor to the request with this Groovy code:

  // Use the request URL as the unique cache key
  def cacheKey = sampler.getUrl().toString()
  // Store the response content in JMeter's global properties
  def cachedResponse = prev.getResponseDataAsString()
  props.put(cacheKey, cachedResponse)
  

• Step 2: Reuse the cache in main threads
  In your main Thread Group, add a JSR223 PreProcessor to every HTTP request that should use the cached content. Use this code:

  def cacheKey = sampler.getUrl().toString()
  if (props.containsKey(cacheKey)) {
      // If cache exists, override the request to return cached data
      sampler.setResponseData(props.get(cacheKey).getBytes())
      sampler.setSuccessful(true)
      sampler.setResponseCode("200")
      sampler.setResponseMessage("Served from global cache")
  }
  

  This checks if the URL is in the global cache first—if yes, it skips sending the real request and returns the cached response instead.

3. Use a Shared File Cache #

You can also configure JMeter to use a single file-based cache that all threads read from:

• First, run a single thread to populate the cache:
  • Add a regular HTTP Cache Manager to a temporary Thread Group, set Cache Storage Implementation to File Cache, and specify a local file path for the cache (e.g., ./jmeter_global_cache).
  • Run this thread once to generate the cache file.
• Then, in your main test plan:
  • Remove the temporary Thread Group, and add the HTTP Cache Manager to your main Thread Group (or root level).
  • Set the same Cache Storage Implementation to File Cache and point it to the same cache file path you used earlier.
  • Enable Use Cache Control/Expires header when processing requests if needed. Now all threads will read from this shared cache file.

Important Notes #

• For the script-based approach, JMeter's props object is thread-safe, so you don't have to worry about race conditions when accessing the cache.
• If your cached content has expiration rules, you'll need to add logic to check and invalidate stale entries (either in the plugin's settings or your custom scripts).
• The plugin method is recommended for most cases since it handles edge cases like cache invalidation and HTTP compliance out of the box.

内容的提问来源于stack exchange，提问作者Jugi