#44 ✓invalid
Alex MacCaw

Reloading Rails Dynamically

Reported by Alex MacCaw | February 13th, 2008 @ 03:32 AM | in Future

I'm not sure if this is a good idea (or possible) and I'm not sure if this is the right forum for it.

However, wouldn't it be nice to deploy without having to restart anything. Is it not possible to catch a HUP signal in Thin to trigger reloading of the Rails app? I think the ruby 'load' method reloads the file, even if it's already loaded (unlike require). What do you think?

Comments and changes to this ticket

  • macournoyer

    macournoyer February 13th, 2008 @ 12:55 PM

    • Milestone set to Future
    • State changed from “new” to “hold”

    Mongrel does it w/ the HUP signal but it's not working very well.

    As it's written when Mongrel boots:

    Rails signals registered.  HUP => reload (without restart).  It might not work well.
    

    I don't think anybody is using this.

    But if you got a stable way to reload a Rails app, I'd be glad to include it.

  • Alex MacCaw

    Alex MacCaw February 13th, 2008 @ 06:50 PM

    Well, here's the patch. I've also added reload support to the runner, so you can do:

    thin reload --servers 3 -e production

    and it'll go through all of them giving them the HUP signal.

    I'll do a bit more testing to see how stable it is (it's working fine at the moment though). Basically I'm doing exactly what the console's reload! method does, but with a few added extras that ensure controllers (and hopefully plugins) get reloaded too. Maybe it would be a good idea to catch a reload exception (if any) and restart instead?

  • macournoyer

    macournoyer February 13th, 2008 @ 08:54 PM

    • State changed from “hold” to “open”

    That definitely require some testing before including. But it looks very interesting.

    thx for the patch Alex

  • Alex MacCaw

    Alex MacCaw February 14th, 2008 @ 01:32 PM

    Unfortunately, a colleague from work tells me that it's likely to suffer from memory leaks, even if you kick of the garbage collector manually. So restarting is the only sure way of cleaning up. I suppose the other option would be a rolling restart, perhaps that could be added to thin's cluster support?

  • macournoyer
  • Alex MacCaw

    Alex MacCaw February 14th, 2008 @ 06:17 PM

    Yes, we could have an options to do a simple rolling restart, i.e. making sure only one thin server is down at any time. Would you like a patch for that?

  • macournoyer

    macournoyer March 16th, 2009 @ 09:03 PM

    • Tag set to rails
    • State changed from “open” to “invalid”

    create another ticket w/ patch for rolling restart as an option

Please Sign in or create a free account to add a new ticket.

With your very own profile, you can contribute to projects, track your activity, watch tickets, receive and update tickets through your email and much more.

New-ticket Create new ticket

Create your profile

Help contribute to this project by taking a few moments to create your personal profile. Create your profile ยป

People watching this ticket

Attachments

Tags

Pages