Page MenuHomePhabricator

Migration of pt.wikimedia.org
Closed, ResolvedPublic

Description

Author: nunotavares

Description:
We at Wikimedia PT would like to migrate pt.wikimedia.org contents to an external server. Is it possible to provide a dump somewhere with current site contents?

After migration, is it possible to redirect pt.wikimedia.org to the new site for a while?


Version: unspecified
Severity: normal

Details

Reference
bz23537

Event Timeline

bzimport raised the priority of this task from to Low.Nov 21 2014, 11:03 PM
bzimport set Reference to bz23537.
bzimport added a subscriber: Unknown Object (MLST).

jeluf wrote:

Do you need a SQL or XML dump?

I fear that I can't provide you with a dump of the user database, for privacy reasons and due to the SUL configuration.

It would be possible to point pt.wikimedia.org to your server or to send a redirect to another URL, if you prefer that.

nunotavares wrote:

Jeluf, we'd like to have a SQL dump, if that's possible. If you need an external location to put it online just drop me a private note to send you credentials.

There's no problem at all with the user table, we'll manage to keep same usernames.

As soon as we reinstate the website on our site, we'd much appreciate you guys setup the redirect as soon as possible. Is there a faster way for getting the redirect done than this bugzilla?

(In reply to comment #2)

Jeluf, we'd like to have a SQL dump, if that's possible. If you need an
external location to put it online just drop me a private note to send you
credentials.

There's no problem at all with the user table, we'll manage to keep same
usernames.

The privacy issues that he is referring to issues such as the users IP addresses.

nunotavares wrote:

I'm not sure I understand what you are referring to.

I already said we don't care about the user table. Last time I looked at mediawiki schema, the only user data which is retained on a SQL dump is username/IP repetition [which, since the the wiki is open, it's already public] so, and having already said that, I don't understand what you mean.

I'm somewhat not sure that redirecting *.wikimedia.org domains outside our cluster is a good idea (cookies and whatnot).

It's ok, we can survive without the redirect. :)

I doubt anyone was typing that domain directly anyway, since we always advertised the domain wikimedia.pt instead, which currently redirects there and is easier to remember and type.

well, I mean, we can do http redirect to wikimedia.pt, DNS record pointing outside cluster would be evil, though

Abigor wrote:

its much faster to export all the content using [[special:export]] because when you miss all the user tables this will cause sql errors and the site will not function very good.

When the wiki isn't that big like yours I would just use special:Export or ask a xml dump from the content, there is no need to get yourself in lots of trouble because you miss tables.

I created a xml dump for all the pages and talkpages, but excluded the '''file''' and '''user''' namespace.
http://www.wikiweet.nl/dump/

nunotavares wrote:

Hi Domas, thanks for pointing that out. I was actually requesting an HTTP redirect, indeed. Sorry not being more clear. "Location: " is enough.

In the meantime, I developed a script to fetch all images based on imagelinks, so no problem with images - which also means XML dump is good enough. I also don't know how to do a full export with Special:Export, but it seems you managed, so I'm downloading it right away.

I'll let you know when we're done.

Thanks for everything.

Just pointing out a detail that was mentioned on IRC: if the redirect could work dynamically (i.e. to have deep links such as diffs, etc redirected to the corresponding url in the new domain) that would be ideal. That is, something like what currently we get if we type an url without the /wiki/ part.

This would prevent breaking links (from mailing list archives and elsewhere).

Its been a few months and no activity on this bug. Still no redirect,
but wikimedia.pt looks like it is up and in use. If there is nothing
else wrong right now, then all that is left is setting up a redirect.

I'm going to close this bug. If the only thing left is the redirect
then that is Bug #30123.