Add support for sending exceptions to Sentry

[cdrx]

[TonsOfFun]

Could we use the Worker._fail to send notifications to error monitors.

[TonsOfFun]

Something like this:

import honeybadger
from faktory import Worker

# Configure Honeybadger
honeybadger.configure(api_key = 'your_api_key')

class HoneybadgerWorker(Worker):
    def _fail(self, jid: str, exception=None):
        response = {"jid": jid}
        if exception is not None:
            response["errtype"] = type(exception).__name__
            response["message"] = str(exception)
            # notify Honeybadger of the exception
            honeybadger.context(jid=jid)
            honeybadger.notify(exception)

        self.faktory.reply("FAIL", response)
        ok = next(self.faktory.get_message())

# Now, when you create a new worker class, inherit from HoneybadgerWorker instead of Worker

class MyWorker(HoneybadgerWorker):
    # Your worker code goes here
    pass

[cdrx]

What you suggested is roughly how I would implement it; only I wouldn't duplicate the existing worker code. Something like this is safer:

    def _fail(self, jid: str, exception=None):
        super()._fail(jid, exception)
        if exception is not None:
            honeybadger.context(jid=jid)
            honeybadger.notify(exception)

[cdrx]

One other thing. I have no idea what honeybadger is, but if you are using threads then this does not look safe to me:

honeybadger.context(jid=jid)
honeybadger.notify(exception)

Feels a little sketchy.

[TonsOfFun]

Yeah I was wondering about that, in terms of overriding the _fail behavior. Honeybadger is an error monitoring provider, but not sure on thread safety in this case. I'll have to test it out, thanks for the input!