Hello, I apologize if I was not clear enough. The dumps are being run manually because Ariel is currently testing the new framework. It is thus to be expected if the dumps are not being produced according to schedule. Also, I believe the enwiki dump is still being produced, so I doubt that the process is "dead", it's just slow. Perhaps Ariel is able to provide more information on this issue, as I am unable to comment further than this.

Probably no one else does know (how to run these dumps). So to clear up a little of the confusion/lack of information, automating staged dumps is currently tracked in Phabricator here: https://phabricator.wikimedia.org/T107750 Just got the configuration files puppetized for the scheduler, which was started manually this time but won't need any other intervention. I'll be keeping an eye on it anyways since this too is new code. I've been playing whack-a-mole with bugs in the streaming stubs/logs production, which is why the delay again last month in completing enwp and why only one set of dumps went off. This run should be better, cross fingers that nothing else crept in. After one complete run finishes for the non en wikipedia dumps, we'll be able to do a second run, probably leaving off revision history since most folks don't need those more than once a month. <-- If this is wrong, speak up now. The order in which things are run has been redone a bit, hopefully to get the parts of the dumps people need done sooner in the cycle. I'm still reluctant to do partial dump copies over labs unless we can guarantee a little more space over there to keep e.g. the last full good dump along with whatever new files are being copied. I've opened a Phabricator task for that here https://phabricator.wikimedia.org/T108077 and added Coren so we can figure out the space issue. I've not wanted to weigh in earlier until I had things in place to kick off the new run. My apologies if that just made folks more annoyed. Ariel Στις 04-08-2015, ημέρα Τρι, και ώρα 20:08 -0600, ο/η Bryan White έγραψε: