One neat use case for the HTTP ETag header is client-side HTTP caching for GET requests. Along with the ETag header, the caching workflow requires you to fiddle with other conditional HTTP headers like If-Match or If-None-Match. However, their interaction can feel a bit confusing at times.

Every time I need to tackle this, I end up spending some time browsing through the relevant MDN docs123 to jog my memory. At this point, I’ve done it enough times to justify spending the time to write this.

Caching the response of a GET endpoint

The basic workflow goes as follows:

  • The client makes a GET request to the server.
  • The server responds with a 200 OK status, including the content requested and an ETag header.
  • The client caches the response and the ETag value.
  • For subsequent requests to the same resource, the client includes the If-None-Match header with the ETag value it has cached.
  • The server regenerates the ETag independently and checks if the ETag value sent by the client matches the generated one.
    • If they match, the server responds with a 304 Not Modified status, indicating that the client’s cached version is still valid, and the client serves the resource from the cache.
    • If they don’t match, the server responds with a 200 OK status, including the new content and a new ETag header, prompting the client to update its cache.
Client                                 Server
  |                                       |
  |----- GET Request -------------------->|
  |                                       |
  |<---- Response 200 OK + ETag ----------|
  |     (Cache response locally)          |
  |                                       |
  |----- GET Request + If-None-Match ---->|  (If-None-Match == previous ETag)
  |                                       |
  |       Does ETag match?                |
  |<---- Yes: 304 Not Modified -----------|  (No body sent; Use local cache)
  |       No: 200 OK + New ETag ----------|  (Update cached response)
  |                                       |

We can test this workflow with GitHub’s REST API suite via the GitHub CLI4. If you’ve installed the CLI and authenticated yourself, you can make a request like this:

gh api -i  /users/rednafi

This asks for the data associated with the user rednafi. The response looks as follows:

HTTP/2.0 200 OK
Etag: W/"b8fdfabd59aed6e0e602dd140c0a0ff48a665cac791dede458c5109bf4bf9463"


I’ve truncated the response body and omitted the headers that aren’t relevant to this discussion. You can see that the HTTP status code is 200 OK and the server has included an ETag header.

The W/ prefix indicates that a weak validator5 is used to validate the content of the cache. Using a weak validator means when the server compares the response payload to generate the hash, it doesn’t do it bit-by-bit. So, if your response is JSON, then changing the format of the JSON won’t change the value of the ETag header since two JSON payloads with the same content but with different formatting are semantically the same thing.

Let’s see what happens if we make the same request again while passing the value of the ETag in the If-None-Match header.

gh api -i -H \
    'If-None-Match: W/"b8fdfabd59aed6e0e602dd140c0a0ff48a665cac791dede458c5109bf4bf9463"' \

This returns:

HTTP/2.0 304 Not Modified
Etag: "b8fdfabd59aed6e0e602dd140c0a0ff48a665cac791dede458c5109bf4bf9463"

gh: HTTP 304

This means that the cached response in the client is still valid and it doesn’t need to refetch that from the server. So, the client can be coded to serve the previously cached data to the users when asked for.

A few key points to keep in mind:

  • Always wrap your ETag values in double quotes when sending them with the If-None-Match header, just as the spec says6.

  • Using the If-None-Match header to pass the ETag value means that the client request is considered successful when the ETag value from the client doesn’t match that of the server. When the values match, the server will return 304 Not Modified with no body.

  • If we’re writing a compliant server, when it comes to If-None-Match, the spec tells us7 to use a weak comparison for ETags. This means that the client will still be able to validate the cache with weak ETags, even if there have been slight changes to the representation of the data.

  • If the client is a browser, it’ll automatically manage the cache and send conditional requests without any extra work.

Writing a server that enables client-side caching

If you’re serving static content, you can configure your load balancer to enable this caching workflow. But for dynamic GET requests, the server needs to do a bit more work to allow client-side caching.

Here’s a simple server in Go that enables the above workflow for a dynamic GET request:

package main

import (

// calculateETag generates a weak ETag by SHA-256-hashing the content
// and prefixing it with W/ to indicate a weak comparison
func calculateETag(content string) string {
    hasher := sha256.New()
    hash := hex.EncodeToString(hasher.Sum(nil))
    return fmt.Sprintf("W/\"%s\"", hash)

func main() {
    http.HandleFunc("/", func(w http.ResponseWriter, r *http.Request) {
        // Define the content within the handler
        content := `{"message": "Hello, world!"}`
        eTag := calculateETag(content)

        // Remove quotes and W/ prefix for If-None-Match header comparison
        ifNoneMatch := strings.TrimPrefix(
            strings.Trim(r.Header.Get("If-None-Match"), "\""), "W/")

        // Generate a hash of the content without the W/ prefix for comparison
        contentHash := strings.TrimPrefix(eTag, "W/")

        // Check if the ETag matches; if so, return 304 Not Modified
        if ifNoneMatch == strings.Trim(contentHash, "\"") {

        // If ETag does not match, return the content and the ETag
        w.Header().Set("ETag", eTag) // Send weak ETag
        w.Header().Set("Content-Type", "application/json")
        fmt.Fprint(w, content)

    fmt.Println("Server is running on http://localhost:8080")
    http.ListenAndServe(":8080", nil)
  • The server generates a weak ETag for its content by creating a SHA-256 hash and adding W/ to the front, indicating it’s meant for weak comparison.

    You could make the calculateETag function format-agnostic, so the hash stays the same if the JSON format changes but the content does not. The current calculateETag implementation is susceptible to format changes, and I kept it that way to keep the code shorter.

  • When delivering content, the server includes this weak ETag in the response headers, allowing clients to cache the content along with the ETag.

  • For subsequent requests, the server checks if the client has sent an ETag in the If-None-Match header and weakly compares it with the current content’s ETag by independently generating the hash.

  • If the ETags match, indicating no significant content change, the server replies with a 304 Not Modified status. Otherwise, it sends the content again with a 200 OK status and updates the ETag. When this happens, the client knows that the existing cache is still warm and can be served without any changes to it.

You can spin up the server by running go run main.go and from a different console, start making requests to it like this:

curl -i  http://localhost:8080/foo

This will return the ETag header along with the JSON response:

HTTP/1.1 200 OK
Content-Type: application/json
Etag: W/"1d3b4242cc9039faa663d7ca51a25798e91fbf7675c9007c2b0470b72c2ed2f3"
Date: Wed, 10 Apr 2024 15:54:33 GMT
Content-Length: 28

{"message": "Hello, world!"}

Now, you can make another request with the value of

the ETag in the If-None-Match header:

curl -i -H \
    'If-None-Match: "1d3b4242cc9039faa663d7ca51a25798e91fbf7675c9007c2b0470b72c2ed2f3"' \

This will return a 304 Not Modified response with no body:

HTTP/1.1 304 Not Modified
Date: Wed, 10 Apr 2024 15:57:25 GMT

In a real-life scenario, you’ll probably factor out the caching part in middleware so that all of your HTTP GET requests can be cached from the client-side without repetition.

One thing to look out for

While writing a cache-enabled server, make sure the system is set up so that the server always sends back the same ETag for the same content, even when there are multiple servers working behind a load balancer. If these servers give out different ETags for the same content, it can mess up how clients cache that content.

Clients use ETags to decide if content has changed. If the ETag value hasn’t changed, they know the content is the same and don’t download it again, saving bandwidth and speeding up access. But if ETags are inconsistent across servers, clients might download content they already have, wasting bandwidth and slowing things down.

This inconsistency also means servers end up dealing with more requests for content that clients could have just used from their cache if ETags were consistent.

Recent posts

  • Protobuffed contracts
  • TypeIs does what I thought TypeGuard would do in Python
  • Crossing the CORS crossroad
  • Dysfunctional options pattern in Go
  • Einstellung effect
  • Strategy pattern in Go
  • Anemic stack traces in Go
  • Retry function in Go
  • Type assertion vs type switches in Go
  • Patching pydantic settings in pytest