How to set up Varnish caching for GraphQL in Django - django

Question

Here is my django-benchmark project in which I implemented simple REST API and GraphQL endpoints. In front of my application I put Varnish for caching. Caching works well for Rest HTTP endpoints and does not work for GraphQL here is my Varnish configuration. What am I doing wrong?
vcl 4.1;
# Default backend definition. Set this to point to your content server.
backend default {
.host = "0.0.0.0";
.port = "8080";
}
sub vcl_hash {
# For multi site configurations to not cache each other's content
if (req.http.host) {
hash_data(req.http.host);
} else {
hash_data(server.ip);
}
# Cache GET requests coming to endpoint located at /api/graphql
if (req.method == "POST" && req.url ~ "/api/graphql") {
call process_graphql_headers;
}
}
# TODO: Find way to cache GraphQL requests.
sub process_graphql_headers {
}
sub vcl_recv {
# # Bypass authenticated requests, these requests should not be cached by default
# if (req.http.Authorization ~ "^Bearer") {
# return (pass);
# }
}
sub vcl_backend_response {
}
sub vcl_deliver {
}
Application Setup
Set up the project and seed database.
$ git clone https://github.com/ldynia/django-benchmark.git
$ cd django-benchmark/
$ docker-compose up -d
$ docker exec -it django-benchmark python manage.py seed 100
Application Testing
Query graphql endpoint directly (port 8080), or passing by varnish first (port 8888).
# Query graphql endpoint directly
$ curl 'http://localhost:8080/api/graphql/' \
-X 'POST' \
-H 'Content-Type: application/json' \
--data-raw '{"query":"query { allDummy { results { id } }}"}'
# Query graphql endpoint hitting varnish first
$ curl 'http://localhost:8888/api/graphql/' \
-X 'POST' \
-H 'Content-Type: application/json' \
--data-raw '{"query":"query { allDummy { results { id } }}"}'
Response
{
"data": {
"allDummy": {
"results": [{
"id": "1"
}, {
"id": "2"
}, {
"id": "3"
},
...
]
}
}
}
HTTP Request & Response Headers
$ curl 'http://localhost:8888/api/graphql/' -X 'POST' -H 'Content-Type: application/json' --data-raw '{"query":"query { allDummy { results { id } }}"}' -v > /dev/null
Note: Unnecessary use of -X or --request, POST is already inferred.
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0* Trying 127.0.0.1:8888...
* TCP_NODELAY set
* Connected to localhost (127.0.0.1) port 8888 (#0)
> POST /api/graphql/ HTTP/1.1
> Host: localhost:8888
> User-Agent: curl/7.68.0
> Accept: */*
> Content-Type: application/json
> Content-Length: 48
>
} [48 bytes data]
* upload completely sent off: 48 out of 48 bytes
* Mark bundle as not supporting multiuse
< HTTP/1.1 200 OK
< Date: Mon, 08 Mar 2021 13:45:42 GMT
< Server: WSGIServer/0.2 CPython/3.7.9
< Content-Type: application/json
< Vary: Cookie
< X-Frame-Options: DENY
< Content-Length: 17128
< X-Content-Type-Options: nosniff
< Referrer-Policy: same-origin
< Set-Cookie: csrftoken=86bgA2o83BavIOTq7Wf59pXxZPeJ65byTMt286UbyKfPSo9O1uefGw8gMP99plbL; expires=Mon, 07 Mar 2022 13:45:42 GMT; Max-Age=31449600; Path=/; SameSite=Lax
< X-Varnish: 32791
< Age: 0
< Via: 1.1 varnish (Varnish/6.4)
< Accept-Ranges: bytes
< Connection: keep-alive
<
{ [17128 bytes data]
100 17176 100 17128 100 48 124k 358 --:--:-- --:--:-- --:--:-- 125k