For /media/ changes, generate a classic purge based on the URL. For
anything tha'ts not templates *or* media, just don't generate any purges
at all (of course).
It seems the newer tidy in Debian Stretch breaks with the output from
the old docs toolchain, causing indention to happen inside <pre> blocks
which clearly breaks rendering.
Turn it off for thos, but keep it enabled for version 11 and up (at this
point that's just dev), because the output becomes a lot easier to read
when trying to debug things.
The code was supposed to pick the latest version of the repo rpm in this
scenario, but at some point the data structure was updated and the code
was not, so it ended up actually picking the *oldest* version of the
repo rpm.
This callback can prevent new users from being created based on specific
criteria. The plugin is defined in settings.py, so it can be any python
code. For example, it can look up in a specific table if this user is
supposed to be allowed to log in at all.
This sample is used across most of our other sites, so add functionality
to install without having a local master repo.
Back-patch of pgeu commit 3e4d8c9101c2d309b134c2f5638aa40db8ef2cf5
There were per-site configured rules defined but the regexp was slightly
incorrectly defined. However, we should just simply never crawl urls
like this unless they are normalized, so for now just add them to the
hardcoded exclusion rules.
We do this by always spidering every repo rpm available (since the ftp
server doesn't know about supported versions), and then filtering which
versions are shown in the dropdown instead. Given the small number of
versions of either kind, this doesn't make a material difference in the
size of the files anyway.
The ftp server can then submit a list (and structure) of which platforms
are supported for yum downloads, which can then later (in a separate
commit) be used to generate a nicer download for yum repo rpms.
The main reason this didn't work before was because of server dictionary
configuration, not because of the code. But make it explicit so we're
not relying on session level settings.
We only support it for our main website, which uses a sitemap, so
implement it only for that provider. And always probe
sitemap_internal.xml, since we don't even try to access any external
sites on it.
Uncommented "add length 7" to array deindex seems to have come from it
being the length of http://. Now changed to https://, so change the
length as well, and properly comment it.
The queue used for varnish purges has so few entries that it's really
not worth paying the management overhead for skytools/pgq. Instead we
can use a very simple local deamon using LISTEN/NOTIFY to fire them off.
Now include a proper nagios plugin in this package, so we can get rid
of the not-very-nice munin plugin currently used in the deployment.
This removes the dependency on django from docload, facilitating
incremental upgrades of the infrastructure.
This now requires a new docload.ini file in the tools/docs directory,
with a section "db" and a setting "dsn".
This has been in standard python since 2.6, and we don't really care
about supporting anything older than that (even our old and soon to
be upgraded environments use 2.7)
