Django Cache-control header: Why isn't Chrome caching this resource? - django

I have decorated a Django view with cache_control as follows:
#cache_control(
private=True,
max_age=5 * 60, # 5 minutes
)
def my_view(req):
…
When I try it with the local test server, it works as expected: subsequent page views in Chrome use the cached resource and don't make a request. When deployed in production, though, Chrome seems to ignore the Cache-control header and makes a new request every time I hit that page.
Here's the full list of headers that the production server responds with:
Cache-Control:private, max-age=300
Connection:close
Content-Encoding:gzip
Content-Length:13135
Content-Type:text/html; charset=utf-8
Date:Wed, 22 Jan 2014 20:39:29 GMT
P3P:CP="IDC CURa ADMa OUR IND PHY ONL COM STA"
Server:nginx/1.4.1
Set-Cookie:csrftoken=87y26bT5uPmyA9wt51N7m4blyqBH5nSo; expires=Wed, 21-Jan-2015 20:39:29 GMT; Max-Age=31449600; Path=/
Vary:Cookie,Accept-Encoding
What could be going wrong? Any ideas? Thanks in advance!

Got it: it was a combination of Google Analytics' cookie and the Vary:Cookie header (set by Django's SessionMiddleware). Analytics' cookie changes with each request, but since ga.js doesn't load when working on localhost, the problem only showed up in production.

Related

Django+Angular CORS not working with POST

My Angular4 app (running on http://127.0.0.1:4200 development server) is supposed to access a django REST backend on the web. The backend is under my control and is available only via HTTPS (running Apache that tunnels the request to a gunicorn server running on an internal port). Let's say that this is https://example.com/. For historical reasons, logging the user in is done using sessions, because I want the users to be able to also use Django's admin interface after they logged in. The workflow is as follows:
Users opens http://127.0.0.1:4200, I perform a GET request to https://example.com/REST/is_logged_in which returns a 403 when the user isn't logged in via sessions yet, 200 otherwise. In the former case, the user is redirected to https://example.com/login/, rendered by Django's template engine, allowing the user to log in. Once logged in, the user is redirected to http://127.0.0.1:4200
When clicking on some button in my Angular UI, a POST request is performed. This post request fails with 403, even though the preflight OPTIONS request explicitly lists POST as allowed actions.
Here is my CORS configuration in Django:
NG_APP_ABSOLUTE_URL = 'http://127.0.0.1:4200'
# adapt Django's to Angular's presumed XSRF cookie/header names
CSRF_COOKIE_NAME = "XSRF-TOKEN"
CSRF_HEADER_NAME = "HTTP_X_XSRF_TOKEN"
CORS_ORIGIN_WHITELIST = (
urlparse(NG_APP_ABSOLUTE_URL).netloc
)
CSRF_TRUSTED_ORIGINS = (
urlparse(NG_APP_ABSOLUTE_URL).netloc
)
CORS_ALLOW_HEADERS = default_headers + (
'x-xsrf-token',
)
CORS_ALLOW_CREDENTIALS = True
This is what Chrome reports for the (successful, 200) first REST GET request to check whether the user is logged in (after he successfully did) in the response:
Access-Control-Allow-Credentials:true
Access-Control-Allow-Origin:http://127.0.0.1:4200
Allow:GET, HEAD, OPTIONS
Connection:close
Content-Type:application/json
Date:Wed, 26 Apr 2017 15:09:26 GMT
Server:gunicorn/19.6.0
Set-Cookie:XSRF-TOKEN=...; expires=Wed, 25-Apr-2018 15:09:26 GMT; Max-Age=31449600; Path=/
Transfer-Encoding:chunked
Vary:Accept,Cookie,Origin
X-Frame-Options:SAMEORIGIN
The corresponding request had this:
Cookie:sessionid=...; XSRF-TOKEN=...
Host:example.com
Origin:http://127.0.0.1:4200
Referer:http://127.0.0.1:4200/
Now, to the actual problem:
Preflight request:
Request URL:https://example.com/REST/change_user_data/
Request Method:OPTIONS
Status Code:200 OK
Access-Control-Request-Headers:content-type
Access-Control-Request-Method:POST
Connection:keep-alive
Host:example.com
Origin:http://127.0.0.1:4200
Referer:http://127.0.0.1:4200/dashboard/account
Preflight response:
Access-Control-Allow-Credentials:true
Access-Control-Allow-Headers:accept, accept-encoding, authorization, content-type, dnt, origin, user-agent, x-csrftoken, x-requested-with, x-xsrf-token
Access-Control-Allow-Methods:DELETE, GET, OPTIONS, PATCH, POST, PUT
Access-Control-Allow-Origin:http://127.0.0.1:4200
Access-Control-Max-Age:86400
Connection:close
Content-Length:0
Content-Type:text/html; charset=utf-8
Date:Wed, 26 Apr 2017 15:36:56 GMT
Server:gunicorn/19.6.0
Vary:Origin
X-Frame-Options:SAMEORIGIN
Now my failing (403) POST request:
Accept:application/json
Accept-Encoding:gzip, deflate, br
Accept-Language:de-DE,de;q=0.8,en-US;q=0.6,en;q=0.4
Connection:keep-alive
Content-Length:60
Content-Type:application/json
Cookie:sessionid=...; XSRF-TOKEN=...
Host:example.com
Origin:http://127.0.0.1:4200
Referer:http://127.0.0.1:4200/dashboard/account
The response headers:
HTTP/1.1 403 Forbidden
Date: Wed, 26 Apr 2017 15:36:56 GMT
Server: gunicorn/19.6.0
Vary: Accept,Cookie,Origin
X-Frame-Options: SAMEORIGIN
Content-Type: application/json
Access-Control-Allow-Credentials: true
Allow: POST, OPTIONS
Access-Control-Allow-Origin: http://127.0.0.1:4200
Set-Cookie: XSRF-TOKEN=...; expires=Wed, 25-Apr-2018 15:36:56 GMT; Max-Age=31449600; Path=/
Connection: close
Transfer-Encoding: chunked
Why wouldn't this request work? It makes little sense to me!
Best regards!
I had the same problem, trying to send a POST request to Django (port 8000) from my Angular CLI (port 4200). I thought it was a problem of Django so I installed cors package however the "problem" is with the browser (actually is not a problem, it is a security issue, see here). Anyway, I solved the problem adding a proxy rule for my Angular CLI, as follows:
First, instead of sending my requests to http://localhost:8000/api/... is send them to /api/ (i.e. to my ng server running at port 4200).
Then I added a file in my Angular project called "proxy.conf.json" with the following content:
{
"/api": {
"target": "http://localhost:8000",
"secure": false
}
}
Finally, run your ng server with the flag "--proxy-config":
ng serve --watch --proxy-config proxy.conf.json
All API requests will be sent to the port 4200 and Angular will internally redirect them to Django, avoiding the CORS problem.
Note that this is only valid for development and won't be used when you build your app code and add it as the static code of your Django server.
Finally, with this solution I didn't need anymore the python module for cors so you could remove it.

Worklight HTTP adapter responseheader cookie

Environment: Worklight 6.2
HTTP Adapter
Response header from backend contains cookies with key (WC_PERSISTENT)
adapter.xml has
<connectionPolicy xsi:type="http:HTTPConnectionPolicyType"
cookiePolicy="IGNORE_COOKIES" maxRedirects="20">
But responseHeaders contains duplicate keys, it seems Worklight server is adding this key.
WC_PERSISTENT cookie is used to track session management in websphere commerce and session timeout is managed in WAS console.
I am new to Websphere, how can I remove the duplicate cookie or not let Worklight add it in responseHeader?
Sample cookie:
"arr": [
"WC_PERSISTENT=pLeLrvmT3fn9TbLo7qEiK1Q5Ss0%3d%0a%3b2015%2d09%2d23+04%3a39%3a39%2e966%5f1442983179966%2d465081%5f0; Expires=Thu, 22 Sep 2016 04:39:39 GMT; Path=\/; Domain=.xyz.com",
"COSTEXCLVAT=disabled; Path=\/; Domain=.xyz.com",
"XYZStats.trailingTag=XYZStats.member,private|; Path=\/; Domain=.xyz.com",
"WC_SESSION_ESTABLISHED=true; Path=\/; Domain=.xyz.com",
"WC_PERSISTENT=kIPpdZQy9v8DUCpCLMHSg6zpkYM%3d%0a%3b2015%2d09%2d23+04%3a39%3a40%2e057%5f1442983179966%2d465081%5f7; Expires=Thu, 22 Sep 2016 04:39:39 GMT; Path=\/; Domain=.xyz.com",
"WC_AUTHENTICATION_1632733901=1632733901%2cuIeVQv0MRMpXTg32wGaT9zj1QVM%3d; Path=\/; Domain=.xyz.com; Secure",
"WC_ACTIVEPOINTER=%2d20%2c7; Path=\/; Domain=.xyz.com",
"WC_USERACTIVITY_16327111111=163211111%2c7%2c0%2cnull%2cnull%2cnull%2cnull%2cnull%2cnull%2cnull%2cbUjUqwUJf8g6D2mrgAw%2fkNzqKJpY%2fs4MXGXY53%2fpMwMxcerOUca5SRGOS%2fmVYKga2zpr9CpT57L9%0akK%2fF3nRmN47DSCdUKGitR4vnxvU%2brBuoX%2f0vLXPmVYLT1XNkgPfYhXPpAm33UHHDpgqqArBnZg%3d%3d; Path=\/; Domain=.xyz.com"
]
But if we request back-end from curl we get only 1 instance of WC_PERSISTENT
From a Worklight perspective it does not duplicate these cookies. It does sound like the cookies come from WebSphere Commerce ("WC"). I do suggest to contact the WebSphere Commerce support team.
Also note that setting the cookiePolicy of the HTTP adapter as "ignoreCookies" does not mean that cookies will not be added - again, these cookies are coming from the response from your backend/application server, not Worklight. Read more about an HTTP adapter's cookiePolicy parameter here: What does the cookiePolicy adapter parameter do in Worklight 6.1+?

django-paypal: IPN requests are always INVALID

I'm using dcramer's fork of django-paypal, but I always encounter an invalid IPN while working with my sandbox accounts.
I receive the following IPN:
Invalid postback. (INVALID)
I tried everything that showed up on google:
checked seller & buyer emails
sandbox accounts are both verified
I use form.sandbox to render the paypal form
tried removing custom values
there is no non-ascii character in the request
When manually checking the request with https://www.sandbox.paypal.com/cgi-bin/webscr, I also get INVALID.
Did someone encounter this issue ? Is there any more-verbose page to validate ipn requests ?
Yes, I also get errors on post-back starting yesterday (18 June):
Opened POST Back Socket to PayPal.
PayPal Post Back returns HTTP/1.0 400 Bad Request
Server: AkamaiGHost
Mime-Version: 1.0
Content-Type: text/html
Content-Length: 216
Expires: Mon, 18 Jun 2012 22:18:00 GMT
Date: Mon, 18 Jun 2012 22:18:00 GMT
Connection: close
<HTML><HEAD>
<TITLE>Invalid URL</TITLE>
</HEAD><BODY>
<H1>Invalid URL</H1>
The requested URL "/cgi-bin/webscr", is invalid.<p>
....
</BODY></HTML>
: not handled.
I use my own IPN integration. It tries to handle all replies from PayPal, which is why I get the last message (: not handled.) I made a package upgrade yesterday, so I'm not quite sure it is a PayPal problem though.

Anonymous URL Level Caching Failure

I want to enable view level caching for anonymous visitor page views. I've turned on the appropriate Middleware (I believe so at least).
MIDDLEWARE_CLASSES = [
'django.middleware.cache.UpdateCacheMiddleware', # This needs to be first https://docs.djangoproject.com/en/dev/topics/cache/#order-of-middleware-classes
'django.middleware.gzip.GZipMiddleware',
'django.middleware.common.CommonMiddleware',
'django.contrib.sessions.middleware.SessionMiddleware',
'django.contrib.auth.middleware.AuthenticationMiddleware',
'django.contrib.messages.middleware.MessageMiddleware',
'pagination.middleware.PaginationMiddleware',
'django.middleware.transaction.TransactionMiddleware',
'waffle.middleware.WaffleMiddleware',
'django.contrib.redirects.middleware.RedirectFallbackMiddleware',
'django.middleware.cache.FetchFromCacheMiddleware',
]
This should automatically set the appropriate HTTP Headers, right? Well it appears that it does.
Cache-Control max-age=600
Content-Encoding gzip
Content-Type text/html; charset=utf-8
Date Wed, 30 Nov 2011 18:46:05 GMT
Expires Wed, 30 Nov 2011 18:56:05 GMT
Last-Modified Wed, 30 Nov 2011 18:46:05 GMT
Vary:Cookie, Accept-Encoding
Now, the problem is two fold. First, the asset is taking just about 1.7 seconds to receive, which strikes me as too long for a cached page. Second, when I look at this page running the local django server, I still see numerous MySQL queries in the django toolbar. That REALLY indicates that caching is failing.
In firebug, there is a console tab titled "Cache", which shows the following:
Last Modified Wed Nov 30 2011 13:46:05 GMT-0500 (EST)
Last Fetched Wed Nov 30 2011 13:46:05 GMT-0500 (EST)
Expires Wed Nov 30 2011 13:56:03 GMT-0500 (EST)
Data Size 11547
Fetch Count 17
Device disk
That SEEMS to suggest that caching is working. I'm confused. If caching is in fact failing, is it due to the browser's internal algorithm for Last Modified?
Thanks for any suggestions.
Have you decorated the particular views you want to cache?
https://docs.djangoproject.com/en/dev/topics/cache/#the-per-view-cache
In my local dev server, where I use localhost, it seems that the browser is setting max-age = 0, so there is no caching for pages happening.
Are you using Google Analytics on the page? It adds two cookies that varies on each request, and since you have enabled sessions which adds vary-on-cookie that means each requested page is seen as unique by the caching framework.
The workaround is to strip out the Google Analytics cookies. I found some code on django-snippets that does this.
# Middleware to strip out Google Analytics cookies that mess up caching
import re
class StripCookieMiddleware(object):
strip_re = re.compile(r'(__utm.=.+?(?:; |$))')
def process_request(self, request):
try:
cookie = self.strip_re.sub('', request.META['HTTP_COOKIE'])
request.META['HTTP_COOKIE'] = cookie
except:
pass
Add that first in the middleware list.
Read more about Django caching and its problems here: https://groups.google.com/d/msg/django-developers/EojHkVKxVWc/G7iNJsARF4IJ

Setting a Compact Privacy Policy with Django

How do I set a P3P compact privacy policy from Django so that IE accepts cookies from my site when the security settings are on HIGH - i.e. no cookies accepted unless there's a Compact Privacy Policy.
Cheers
Guy
Middleware is the preferred way to do things like this on an "every request" basis. For instance, here is a simple bit of middleware to add the same (example) P3P header to every response Django generates:
In settings.py:
P3P_COMPACT = 'policyref="http://www.example.com/p3p.xml", CP="NON DSP COR CURa TIA"'
MIDDLEWARE_CLASSES += ('myapp.middleware.P3PHeaderMiddleware',)
In myapp/middleware.py:
from django.conf import settings
class P3PHeaderMiddleware(object):
def process_response(self, request, response):
response['P3P'] = getattr(settings, 'P3P_COMPACT', None)
return response
You could also get a similar effect in a single view by setting the P3P header in the response:
def my_view(request):
response = render_to_response('my_template.html')
response['P3P'] = 'CP="NON DSP COR CURa TIA"'
return response
To expand on the topic a little bit, cookies and headers such as the P3P header are both sent at the same time, as part of the response; in fact, under the hood, cookies are set with another response header. You can see the cookie header using curl:
$ curl --head http://www.google.com/
HTTP/1.1 200 OK
Date: Wed, 13 Jan 2010 00:04:59 GMT
Expires: -1
Cache-Control: private, max-age=0
Content-Type: text/html; charset=ISO-8859-1
Set-Cookie: PREF=ID=d2c09762c479f94e:TM=1263341099:LM=1263341099:S=oJby3NpU4RsRfuYa; expires=Fri, 13-Jan-2012 00:04:59 GMT; path=/; domain=.google.com
Set-Cookie: NID=30=kdKrd5e-u6Xs7cUe3p4eaNDtv6SO88uBL5v6_M1XMTSRmkh7okxrWLOm-l_uZdN37PxQIe4dBlekFFVCpTFXGyIDlUrz1hEwhgVLvXfIik_VeVWGmWzKbA5qu_Zq0sOi; expires=Thu, 15-Jul-2010 00:04:59 GMT; path=/; domain=.google.com; HttpOnly
Server: gws
X-XSS-Protection: 0
Transfer-Encoding: chunked
I don't know terribly much about p3p but I did a little digging and found this:
http://www.w3.org/TR/P3P11/#Well_Known_Location
You put the file at /w3c/p3p.xml
It looks as though p3p policies are similar to robots.txt files.
Additionally you can set p3p headers on all your pages if the robots.txt method isn't the way you want to go. That's a side-note, however, since you want the compact version which I'm assuming is the p3p.xml file.
Hope this helps get you on the right track.