Brion Vibber wrote:
Yep, I'll work this in to the dump process which I'm about to restart this evening with a more even size split. (The way a lot of larger wikis took a long time made it diffcult to get smaller, easier wikis done regularly.)
I'm splitting the dumps to four threads now, for huge/large/medium/small-size wikis each: http://meta.wikimedia.org/wiki/User:Brion_VIBBER/Dump_build_split
Currently just the small ones are running in Tampa while I test that it's all still working properly.
As a note: the langlinks and externallinks tables are now included as sql dumps. Some may find these useful or interesting.
externallinks has an el_index field which swaps the order of the components in the hostname, making it easier to count parent domains if one wants to make some aggregate listings.
-- brion vibber (brion @ pobox.com)