Allow optional delay to watch for immediate errors on deploy

[ELLIOTTCABLE]

Some sorts of obvious and quickly-caught errors that slip past you on the way to deploy, don't throw an exception as soon as you run the start command, but still throw within a (short) predictable period after boot. (For instance, some misconfiguration that's preventing your app from connecting to the database, for completely obvious reasons; but still not throwing until five seconds into boot-up.)

I'd love to see a flag, or simply a longer default, to modify the delay that the local jitsu deploy command waits. That way, we don't get “Nodejitsu ok” when, in fact, it immediately crashed; this also saves me having to open the website to check the log and see what actually went wrong.

[Swaagie]

I like the idea behind this, @jcrugzz would we be able to tell forza to wait a bit longer based on the flag? There could be other usecases as well where someone might want to delay the start a bit longer, like heavy pre-start caching of data (e.g. fetching a lot of stuff from the database)

[jcrugzz]

thinking about it, this does seem possible, it will just be a bit tricky to get right. Currently we wait for the app to listen on a port, once that happens, it gets propagated as a success to our deploy server. This is why if it fails right after this happens, its not caught on the initial deploy.

Let me think about how this would work.

[ELLIOTTCABLE]

Thanks for the quick and thoughtful responses, guys. Just glad to see you find it interesting.

hearts! <3