On Thursday, 1 November 2012 23:50:16 UTC+11, Richard Penman wrote: > > Would be great to use the new scheduler but doesn't seem applicable in > this case. I have no difficulty starting these processes. > > As described in OP, the problem is these are child processes which makes > them dependent on the main web2py process. So I was wanting these to > be independent processes - or is there a better way? > > I've been down this path twice. Both times it ended by using the scheduler. It runs independently of the web2py server. The communication process back to the server is via the database you are using. It works pretty well. Out of the box, the scheduler, which is its own process, spawns workers as necessary to handle the queue of tasks. You seem to be concerned about what happens to tasks if the web2py server process dies. That I'm not sure of, since my jobs don't run across the occasional restarts of the webserver.
--
