Back in May I've got all 26478 questions (from all language sections) backed-up with
wget, after removing JS (embedded and standalone files) they take up ~890MiB of disk
space.
A few dozen questions with multiple pages of answers have only the first page with top
rated answers saved.
I removed all JS with xlstproc, further clean-up of superfluous elements can be done with
it.
Everything is available on
https://askfedora.brtk.098.pl/ but since I only saved
questions, for now you need to go directly to their addresses, e.g. form
https://askfedora.brtk.098.pl/sitemap.xml (robots are asked not to index it).
Links to related questions point to same server, other (AskBot interface links) to
https://askbot.fedoraproject.org/
All that was to backup the most important data with minimal effort. To retain most
usability, either proxy or current Ask Fedora server would have to rewrite domain name and
redirect all requests starting with language codes (e.g.
https://ask.fedoraproject.org/en/question/* https://ask.fedoraproject.org/es/question/*)
to addresses where those html files are hosted. That would open askbot questions from this
backup without interfering with Discourse and breaking old links or search engine results
(well, most of them).
Without working redirects, our next best (ok, acceptable, not best) plan is probably to
add some landing page, further clean-up those files (xlstproc seems powerful, capable of
removing or replacing selected elements) and make sure they are indexed by all search
engines on target domain (probably
askbot.fedoraproject.org, after read-only askbot
expires).
If we were to continue with what I've already got, I'd probably also save user
profile pages, while they are still available. How long until AskBot is shut down?