The check was for is_behind_cache without the (), meaning it always
returned true, which in turn meant we trusted all x-forwarded-for
headers. It was pretty hard to get them into the system, and
we didn't actually use it for anything other than locking survey
submissions, so it's not a big problem.
However, the basic logic was also wrong, as it assumes that all
SSL connections terminate directly at the backend server, which is
not necessarily true anymore.
The new version of the function will trust an X-Forwarded-For as
long as it's set on one of our frontend servers, regardless of if
it's an encrypted connection or not.
When changing an email, we generate a random token and send it to the
new email (the old email is verified by the fact that the user is
already logged in). Once the link in the email is clicked, we update
the actual email.
Basically, user generated email (bug report form) will be sent to the mail
frontends for antispam. Any errors generated there will be ignored and
the mails "dropped on the floor". Other emails keep entering the system
through localhost and delivered there.
Import the code from the PostgreSQL Europe website to handle this, since it's
well proven by now.
Any points that send email now just write them to the database using the
functions in queuedmail.util. This means we can now submit notification
emails and such things within transactions and have them properly roll bcak
if something goes wrong (so no more incorrect notifications when there is
a database error).
These emails are picked up by a cronjob that runs frequently (typically
once per minute or once every 2 minutes) that submits them to the local
mailserver. By doing it out of line, this gives us a much better way of
dealing with cases where mail delivery is really slow.
The submission from the cronjob is now done with smtp to localhost instead
of opening a pipe to the sendmail command - though this should have no
major effects on anything.
This also removes the setting SUPPRESS_NOTIFICATIONS, as no notifications
are actually ever sent unless the cronjob is run. On development systems
they will just go into the queuedmail table, and can be deleted from there.
This allows all models inherited from PgModel to specify which
URLs to purge by either setting a field or defining a function
called purge_urls, at which point they will be purged whenever
the save signal is fired.
Also implements a form under /admin/purge/ that allows for manual
purging. This should probably be extended in the future to show
the status of the pgq slaves, but that will come later.
Includes a SQL function that posts the expires to a pgq queue. For
a local deployment, this can be replaced with a simple void function
to turn off varnish purging.
the fact that when pages are served through Varnish, the request will come
from the Varnish server and not from the client.
Create a /system_information page that shows some information about the
connection to help diagnose how the caches work.
