tapjoy / chore Goto Github PK

Docs do a good job of describing the signals that chore can handle, but there's no mention of how to find the PID of a running chore master. And, since Chore currently doesn't record a pidfile at all, a recommended Upstart or similar configuration would be useful.

Chore 1.5.2 does not seem backwards compatible with earlier versions because the signature of the block passed into #handle_messages has changed.

Out of the box Chore won't work with a Rails app. Job classes in a Rails must be eager loaded in all environments for Chore to work. By default adding it to the eager load path won't work in development, as Rails lies and doesn't REALLY eager load in dev (to allow for class reloading mostly).

So, Chore should document how to setup eager load, and/or document the directory glob require one-liner that's been used in the past.

The documentation mentions a dependency on Dali for SQS consumers, but doesn't mention what that is for. Is it possible to add a bit more explanation as to what that does, and would you be open to a PR replacing memcached with Redis, as I'd rather avoid introducing another piece of infrastructure to our system just for this.

Not sure how much of an issue it actually is, but we handle the queue configurations really inconsistently. In some places it's prefixed, in others its not, in some cases it relies on job classes, in others config. We should unify the queue reference handling, as the current setup is difficult to understand.

If you attempt to specify any of the class based args (consumer, publisher, strategies, etc), you can only use classes that exist inside of Chore. Because Chore doesn't attempt to call Bundler.setup or .require the only classes loaded are Chore itself.

This makes it difficult to configure an app that is using a different queue implementation, without making the application itself aware of the change.

ping @StabbyCutyou