Process exited with status 0 , running COUNT > 1 on heroku worker process

[chrisboulton]

Process exited with status 0 , running COUNT > 1 on heroku worker process

[chrisboulton]

@barock19, if you're still using php-resque on Heroku, are you still experiencing these problems?

[jbrower]

I suspect that this is because the original php process detaches. This leaves nothing for heroku to monitor, so they believe that the worker has crashed. It'd be nice if php-resque didn't deamonize (or had an option not to) when you launch multiple workers at once. That would allow things like this to work better. I hope I made sense, if more explanation is needed, please let me know. :)

[danhunsaker]

I have made some changes to the startup script that seem to be related, especially if @jbrower's comment is accurate. I've changed the forking so that the parent process stays alive and monitors the children it spawned. If any of the children die prematurely for any reason, it spins up a new one to replace it. If, however, it receives a signal to close, it asks all of its children to do the same, then waits for them to shut down before exiting itself.

Is this a setup which would be desired for all users, or just for automated systems such as Heroku or Pagodabox?

[jbrower]

It's something that I would certainly like. :)

[danhunsaker]

Just realized that the version I was developing against is 1.2. It looks as though the changes I've made will still work with 1.3, but I haven't tested how well they do/don't interact with Composer.

[ebernhardson]

@jbrower i ported over the resque-pool gem at one point, which does basically the same. It manages the pool of workers. The docs listed are from the original resque-pool project, but i kept the same functionality. https://github.com/ebernhardson/php-resque-pool

[jbrower]

I think that @danhunsaker might have already developed a solution that's built in and will be making a pull request soon. It does an excellent job of keeping things going and letting you know when something isn't working quite right.

[richardkmiller]

@danhunsaker -- I'm also interested in your patch that keeps the parent alive to monitor the children.

[danhunsaker]

Funny you should mention that. I just now submitted a pull request (#83) for that exact functionality. 😁

[richardkmiller]

Awesome -- looking forward to trying it.