Skip to main content

Django X-Forwarded-For Properly

Project description

The X-Forwarded-For header is used by many reverse proxies to pass the IP addresses of the whole chain of hosts between client and application server. The header looks something like this:


This translates to:

X-Forwarded-For: client, proxy1[, proxy2[...]]

However it is just a header. Most default configurations simply append to the header. It is trivial for a malicious client to deliver a header in the initial request:

X-Forwarded-For: phony, client

What django-xff does

This library provides a decent and configurable middleware to rewrite the request.META['HTTP_REMOTE_ADDR'] to the correct client IP.

This is done by setting a depth of reverse proxies to be trusted alone. The X-Forwarded-For header will additionally be sanitized from any extraneous entries.

By default, if the expected depth of proxies is 3, the client address will be used in all of these examples:

X-Forwarded-For: phony, client, proxy1, proxy2
X-Forwarded-For: client, proxy1, proxy2
X-Forwarded-For: client, proxy


  • Less proxies than expected is allowed by default, for varying lengths of proxy chains, the longest is the only one that can be trusted.

  • No header set is allowed by default and the library does nothing.

What django-xff does not do

This library does not check the IP addresses of any proxies along the path of the message.

This library is unable to detect compromised proxies or any incoming requests that have the right number addresses in the correct header.


  • Separate middleware that checks CIDR for the trusted proxies

  • Separate middleware that checks exact IP addresses for proxies


Add the following to your Django module to enable this middleware for two reverse proxies expected. The middlewares are processed order of appearance. This middleware should go somewhere near the top to avoid giving a potentially malicious user chances to validate passwords with malformed requests:

   <a few middlewares here>
   <more middlewares here>


By default, no attempts are denied. There are several settings to send a 400 (Bad Request) response to failing requests. Strict mode will stop all failing requests:


To prevent only the clearly malicious requests, use the following instead:


To prevent requests that do not come through enough proxies, use the following:


The previous setting implies a Bad Request when there is no X-Forwarded-For header present. The following setting follows the XFF_ALWAYS_PROXY and XFF_STRICT by default but can be set independently:


Even in XFF_LOOSE_UNSAFE mode this will require the header:


For an unsafe setting, in development possibly, you can trust that the first entry is always correct and still get the assumed client IP in the right place, use:


If you want to keep the X-Forwarded-For header untouched even if there are extra entries, use:



In some cases requests from alternate request paths are to be expected. The Amazon Elastic Loadbalancer healthcheck or other administrative tasks need to be available even if they do not match the criteria.

This library accepts URIs as regular expressions to be exempt for checking. These will be exempt for any validation including XFF_STRICT and XFF_HEADER_REQUIRED.

To define the whitelist:


This will allow calling /healthcheck/ and /admin/* from anywhere. It is a daft idea to allow everyone to access the admin site with less requirements than the other parts of the site. For this reason it is possible to respond with 404 (Not Found) when the request arrives through the main entrance:


This will assume that anything below XFF_TRUSTED_PROXY_DEPTH is trusted. The method is naive, but effective.


Dropped requests will be logged. This means that there will be plenty of logs when the library is misconfigured or malicious things are taking place. It is recommended to keep the logs for tracing in case of a real attack. However they can be filtered from development by setting:

    'loggers': {
         'xff.middleware': {
              'handlers': ['null'],
              'propagate': False,

Setting up

It is recommended to enable the middleware with the assumed number of proxies and investigating the logs. If the header is not present or the middleware is not configured, there will be no log entries. If the logs state that the depth is incorrect, it should be reduced. If all requests are considered as spoofing, the depth should probably be increased:



When logs appear correct, control can be increased in increments:




Defining exceptions is feasible with other flags set. The following could be used behind an AWS Elastic Loadbalancer to prevent entry without the proper header set but allow healthcheck to return correctly. The stealth would also mask the same URI with a 404 error:

XFF_EXEMPT_URLS = [r'^health/]

In case there is a chain of reverse proxies, the healthcheck URI is available for all layers except the last one.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

django-xff-1.3.0.tar.gz (23.6 kB view hashes)

Uploaded source

Built Distribution

django_xff-1.3.0-py2.py3-none-any.whl (11.4 kB view hashes)

Uploaded py2 py3

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor NVIDIA NVIDIA PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page