Why not ? It's an intranet app, few users, it does what we need. Its simple
to deploy (one app instead of several separate apps). Much simpler
infrastructure. Now i understand those awesome startups who deploy to
Heroku of AWS or what have you in hopes to build next Facebook. Sure, go
wild, nosql, thousands workers complex infrastructure.
But why would i overcomplicate my small intranet app, just because your
deployment tool is not flexible enough ? It would be simpler for me just
not to use your deployment tool :)
On Friday, May 4, 2012 10:10:32 AM UTC-7, Lyndon wrote:
>
> I don't think that it's generally considered a good idea to have
> long-running processes occurring in-app. I know that on Heroku there
> are 'workers' that you can provision to take care of such tasks. These
> could even run as their own apps that connect to the same resources
> for queueing, etc, in order to reuse the deployment interfaces and
> infrastructure.
>
> Deciding to hold off on a restart of an application due to what would
> usually be asynchronous operations seems a bit fishy.
>
> On Sat, May 5, 2012 at 12:32 AM, Vagif Verdi wrote:
> > I have an intranet app that has a couple of long running operations. Say
> > batch print 50 pdf files to printer. Or submit 10,000 invoices to ADP
> Tax
> > server (end of month).
> >
> > Maybe killing the app server should be optional.
> >
> >
> >
> > On Friday, May 4, 2012 7:55:21 AM UTC-7, Michael Snoyman wrote:
> >>
> >> send the TERM signal to the old app,
>
