Closed Bug 1018727 Opened 9 years ago Closed 9 years ago
system for backfilling translations
We have translation systems now. We need something that can backfill translations for responses for a specific product in a specific range. I think for now we should create an admin page because that's pretty easy to do. If this becomes something we need a bunch of, we can figure out other possibilities later. The admin page would have a form that takes: * start date * end date * product * system You would click on a "backfill" button. It would calculate how many responses it'd be backfilling translations for. You'd click on "backfill-it-now" and it'd generate a bunch of celery tasks just like the post_save mechanisms do.
Putting this in this quarter and making it a P1. We want to have translation system metrics working first. Making this block on that.
Priority: -- → P1
Whiteboard: u=analyzer c=translations p=2 s=input.2014q2
Talked with people using translations and this isn't really needed so we're going to push it off.
Whiteboard: u=analyzer c=translations p=2 s=input.2014q2 → u=analyzer c=translations p=2 s=input.2014q3
Gengo is going down on Sunday and doesn't want anyone hitting their system during that downtime. I poked around the code and there's no good way of shutting off all Gengo API stuff other than going through the products and shutting off auto translation. However, if we shut off auto translation, then the responses we get during that period won't get translated. We have two options at that point: 1. go through and spot translate everything by hand which is hard because the spot translation UX/UI *sucks* (bug #1014874) 2. implement this which would let us shut off auto translation for some period and then later when we turn it back on again, we can backfill translations Seems a lot easier to implement item 2. I think this is a 2 point bugs, so I could probably bang out a "good enough" version in a half day or so. Given that, I'm grabbing it to work on on Monday.
Assignee: nobody → willkg
I pushed this to production earlier and ran a backfilling run. It took long enough that the request timed out. Ricky thinks the part that timed out was between zeus and my browser and that Django actually worked through the entire request. That needs to get figured out. I created bug #1038512 to deal with it. Closing this out now.
Status: NEW → RESOLVED
Closed: 9 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.