Handling a Failed Mongrel Start
December 02, 2007 by Adam Wiggins
We’ve been working our tails off over the past few weeks to process all the feedback you guys have been sending (or that we’ve gleaned from the system logs). I think that this photo of the trashcan under Orion’s desk tells the story pretty well:
He bought that case of Rockstar at Costco last week, and consumed it all as part of our mad dash to squash bugs exposed by our sudden surge of users. Bad for Orion’s health, but good for Heroku’s backend stability. :)
One major area we’ve been dealing with in this past week is the issue of failed mongrel starts. That is, exceptions that occur while the Rails framework is booting, rather than on a page request. These sorts of exceptions often happen with imported apps, because certain types of plugins or gem dependencies may not work out of the box with Heroku yet, or you might just have some odd stuff in your environment.rb that isn’t compatible with the version of Rails we’re running.
Up until recently, this was producing an unhelpful HTTP 502 bad gateway page. Now, you’ll see the contents of your mongrel.log so that you can diagnose the issue.
If you were paying close attention, you might have noticed a “Restart” button that existed for a few days. We’ve removed that in favor of automatically restarting whenever you save a file that needs a restart (like routes.rb or environment.rb). So far this seems to work pretty well – but as always, let us know if you find a case where it doesn’t work as expected.