Platonides wrote:

Felipe Ortega wrote:

...

I hate to be obstinate with this, but according to the results on:

http://download.wikimedia.org/enwiki/20070716/

The 7zip pages-meta-history.xml dump for enwiki is 3.2GB and OK. On 11/06 it was more than 8 GB long, so I think we still have a BIG problem to solve about that...

I hope that someone could be so kind to answer this message, not like the previous one that, apparently, arrived to nowhere...

Thanks.

Felipe.

The storing itself seems ok. No compression errors, not truncated... BUT there're void text revisions: '<text xml:space="preserve" />', probably a lot, but not all. Maybe the system has problems accessing external storage?

I'm refering to enwiki-20070716-pages-meta-history.xml.7z

Thanks for the peek, I'll look at it when I have a chance, which may or may not be before Wikimania.

-- brion vibber (brion @ wikimedia.org)

Brion Vibber wrote:

Brion Vibber wrote:

...

Platonides wrote:

...

The storing itself seems ok. No compression errors, not truncated... BUT there're void text revisions: '<text xml:space="preserve" />', probably a lot, but not all. Maybe the system has problems accessing external storage?

I'm refering to enwiki-20070716-pages-meta-history.xml.7z

Thanks for the peek, I'll look at it when I have a chance, which may or may not be before Wikimania.

Looks like the dump was reading from db3 at the time it went down on July 21.

There's DB error detection and recovery in this part of the dump; it may have gone awry and incorrectly run the rest of the dump output without being able to read in more text -- in this case it should have just bowed out with an error, allowing the dump to get re-run automatically.

I'll see if I can fix it up and start a new run.

Started this a bit ago. Similar errors should now bomb out correctly. :)

-- brion vibber (brion @ wikimedia.org)