I saw this plugin and would have used except it didn't work with the latest version of resque.
I've adapted your concept to work with the latest version. I think this is a bit cleaner and doesn't rely on alias_method_chain and uses the plugin hooks (except for the little bit of instance_eval at the end..)
We added resque-mutli-job-forks to our project today, however, it appears that the workers disappear from the resque-web UI.
When the workers first start they show up in the UI, but after the fork for the second time, they all start to appear to die off. Looking at top the workers are obviously working and the jobs are being processed in the queue.
very nice plugin there. I've tried to track down a problem, where when plugin is enabled, the fork will happen in the child and the child will not exit. E.g. with time of 1 minutes per worker, I will see a fork happening initiallly, but then the next worker which is forked after 1 minute will have the child as parent process and the child will not exit.
Did you have any experiences with the side effecst, e.g. the stats are recorded per process and if breaking out the run loop it calls
unregister_worker (which will probably be a noop since the pid is not recorded any way).
I'm in the need for exactly what your plugin provides!
Although I'm running Resque v1.10.0, not v1.7.1.
I can see in Resque v1.8.0 they depricated process and moved to using perform.
Any objection, if I make the changes for Resque >1.8.0?