I took the site down for a few minutes for a code push. We're trying to fix the "application error" issue, where randomly some requests will take a long time to load and then time out. Previously those timeout errors would clump together and we'd get a period of 20-30 minutes with a 5-10% error rate (this might happen a few times during the day, as our cloud application dynos reset themselves and fail to connect to the database properly). Now, with any luck, we won't get those error clumps and we'll be back down to our baseline timeout rate of about 0.05%. If you do notice it happening, please comment here.

One exception: sometimes we'll push minor code changes without taking the site into maintenance mode, which causes timeout errors for about 30 seconds. I'm looking for a way to reduce those, too.