I e-mailed Chris and Leah a friendly notice last Sunday about these errors and never heard anything back...
I don't think you'll see it fixed any time soon.
Using a fork per request is the easy way out. The problem as you've all seen, is that memory will be used up very very quickly.
The solution is to rewrite the app to use an event loop like select or epoll, or to queue the requests in a message queue: doable, but worth the effort? I doubt it, since the website looks like a hobby rather than a commercial product.
Re:Hurl
by jackowayed on Saturday May 01, 2010 @06:22PM
(#1311336)
Forking could fix the fork problem.
That is, if someone forks hurl (http://github.com/defunkt/hurl) and rewrites it to use one of the fixes you referred to above, I'm sure they'd merge that in and deploy that.
Doing it with Resque would make a lot of sense since hurl uses Redis already.
Well sure, if someone does the work for them.
I'm saying that's probably why Chris and Leah haven't fixed it already and probably for many days to come.
Re:Hurl
by jackowayed on Saturday May 01, 2010 @06:50PM
(#1311384)
Yeah, I know. I just wanted to point out that we're not totally helpless here like we would be with closed-source software, and maybe encourage someone to fix it :)
And I just wanted to point out that this isn't a simple 2-3 line fix so people won't get their hopes up trying to see if it's ready every few hours.
Sounds like a great usecase for Node.js. =)
I assume that the problem is that because they're using Redis, the available memory was filled up with data from the database.
Yep, me too. Seems like a useful service though.
You can check out this video to learn more about hurl http://vimeo.com/6238577