Page MenuHomePhabricator

Wikipedia be-tarask is dumped as be_x_old
Open, Needs TriagePublicBUG REPORT

Description

What happens?:

The be-tarask Wikipedia is dumped as be_x_old:

What should have happened instead?:

The dump should be done under "be_tarask" instead.

Event Timeline

That is quite an oddity. From our listing of all wikis, it is the only one that has such a naming:

select count(1) from wikis;

count(1)
1013
Time taken: 8.368 seconds, Fetched 1 row(s)
select * from wikis where database_code like '%old%';

database_code	domain_name	database_group	language_code	mobile_domain_name	language_name	status	visibility	editability	english_name
be_x_oldwiki	be-tarask.wikipedia.org	wikipedia	be-tarask	be-tarask.m.wikipedia.org	Belarusian (Taraškievica orthography)	open	public	public	Belarusian (Taraškievica) Wikipedia
Time taken: 0.469 seconds, Fetched 1 row(s)

Will ask around why.

It looks like the naming be_x_oldwiki is an historical oddity, and that it would be hard to fix:

That is blocked on T112647: [Task] Investigation: how to handle the rename of a site id in Wikidata, by taking a quick look at the problem, to be honest, renaming a wiki is borderline impossible and extremely hacky specially to make sure nothing breaks during migration.

Since the dumps process doesn't have control over the database naming, I think we will have to live with this. Therefore, I think this ticket is as duplicate of T336156: Make it easier to find backup dumps under unusual names, e.g. be-tarask under be_x_oldwiki.

@Albertvillanovadelmoral LMK if you agree.

So. I got added to the ticket: I want more wiki renames to happen and the path that was decided is this: We never rename the db, that's basically not possible but we will make DNS point to the a different db, that has been done multiple times and to be clear if we wikidata issue gets fixed, we will rename at least ten more wikis so the issue of dumps needs to be fixed or at least communicated to the community.

i.e. We are planning to add more oddities (once a blocker is fixed) so if it's not too hard to fix (e.g. as simple as a redirect in dumps nginx), it should be fixed :) or at least mitigated.

Fair enough @Ladsgroup.

We could tackle this as part of our Dumps 2.0 effort.